JP7612909B2 - Dehydration System - Google Patents

Description

The present invention relates to a dehydration system that separates suspensions such as sludge or slurry into turbid matter and liquid, and in particular to a dehydration system that can be optimally operated automatically.

Conventionally, dehydrators have been used to squeeze suspensions (e.g., sludge) discharged from liquid treatment facilities such as sewage treatment plants, sewage treatment plants, and industrial wastewater treatment plants to separate water from the suspension (i.e., dewater it). In the dewatering process, the suspension to which a coagulant has been injected is stirred in a coagulation tank to form coagulated flocs that facilitate solid-liquid separation, and then dewatered using a concentrator or dehydrator.

One example of a dewatering device is a screw press, known as a sludge dewatering machine. This screw press is equipped with a filter cylinder made of a screen (perforated plate) and a screw placed inside the filter cylinder. By rotating the screw, the sludge put into the filter cylinder is squeezed and dewatered. A back pressure plate that holds back the sludge is placed at the downstream open end of the filter cylinder, and this back pressure plate retains the cake (dewatered sludge) sent by the rotating screw, forming a plug of cake. This plug applies back pressure to the cake sent in later, further squeezing the cake. The cake that forms the plug is pushed by the succeeding cake and gradually discharged from the filter cylinder. In this way, a cake with a low moisture content is formed by the screw press.

JP 2018-111065 A JP 2000-246147 A

By maintaining a low moisture content in the cake that is ultimately obtained from the sludge dewatering process, it is possible to reduce the amount of cake that is subject to disposal, etc. In order to maintain the moisture content of the cake as low and stable as possible, it is important to operate the system in a way that optimizes the condition of the sludge in each process of coagulation and dewatering according to the sludge properties and load. However, the moisture content of the cake can change depending on various factors such as the daily changing sludge properties and the operating conditions of the equipment in each process.

On the other hand, there is an AI (artificial intelligence) controlled dehydration method that makes it possible to achieve fully automated operation by collecting various information inside and outside the sewage treatment facility and using the collected information to set (or reset) an appropriate operating pattern for the dehydration device (see, for example, Patent Document 1). However, even if operation is controlled based on the various collected information, it is difficult to fully grasp the properties of sludge, which is derived from various substances that are discharged as sewage, etc., as a result of different human activities every day. In order to optimize the operation of the dehydration device, it is actually necessary for the operator to visually check the current sludge and to intuitively grasp the condition of the sludge at each process.

However, in order for an operator to optimize the operating conditions based on visual inspection of the sludge, the operator must have a wealth of experience and know-how. Furthermore, the operating conditions may differ depending on the operator, resulting in an unstable moisture content of the cake.

The present invention aims to provide a dewatering system that can obtain the state of suspensions such as sludge and slurry as image data, judge the quality of the operating condition based on the image data, and keep the moisture content of the turbid residue (e.g., cake) within a target range and maintain it stably.

In one aspect, a dehydration system for removing liquid from sludge is provided, the dehydration system comprising: a suspension tank for storing sludge; a flocculation device for stirring the sludge and a flocculant to form a flocculate; a dehydration device for removing liquid from the flocculate to form a cake with a concentration of 3 to 50 wt %; an imaging device for generating image data of the liquid removed from the sludge; and a control system having a trained model constructed by a machine learning algorithm, the control system comprising: a storage device in which the trained model is stored; and a processing device for inputting the image data and status data into the trained model and executing calculations for outputting from the trained model optimal operating parameters for the dehydration system that can keep the moisture content of the cake discharged from the dehydration device within a target range, the status data including the flow rate of the sludge to the flocculation device, the properties of the sludge, and the flow rate of the flocculant.

In one aspect, a dehydration system for removing liquid from sludge is provided, the dehydration system comprising: a suspension tank for storing sludge; a flocculation device for stirring the sludge and a flocculant to form a flocculate; a concentrator for removing liquid from the flocculate to form a concentrate with a concentration of 3 to 15 wt%; a dehydration device for removing liquid from the concentrate to form a cake with a concentration of 3 to 50 wt%; an imaging device for generating image data of the liquid removed from the sludge; and a control system having a trained model constructed by a machine learning algorithm, the control system comprising: a storage device in which the trained model is stored; and a processing device for inputting the image data and status data into the trained model and executing calculations for outputting from the trained model optimal operating parameters for the dehydration system that can keep the moisture content of the cake discharged from the dehydration device within a target range, the status data including the flow rate of the sludge to the flocculation device, the properties of the sludge, and the flow rate of the flocculant.

In one aspect, the image data of liquid removed from the sludge is image data of filtrate discharged from the dewatering device.
In one embodiment, the dewatering device is a screw press or a belt press type dewatering device.
In one aspect, the operating parameters output from the trained model include at least one of the following: the sludge flow rate to the flocculation device being in excess or insufficient relative to the appropriate range; the suitability of the flocculant; the flocculant flow rate being in excess or insufficient relative to the appropriate range; the agitation speed of the flocculation device being in excess or insufficient relative to the appropriate range; and the water flow rate supplied to the flocculation device being in excess or insufficient relative to the appropriate range.
In one aspect, the dehydration device is a screw press having a screw shaft arranged in a filter cylinder and screw blades fixed to the outer surface of the screw shaft, and the operating parameters output from the trained model further include at least one of the rotational speed of the screw shaft being in excess or insufficient relative to an appropriate range, and the rotational torque of the screw shaft being in excess or insufficient relative to an appropriate range.
In one aspect, the dewatering device is a two-stage screw press having a first screw and a second screw arranged in a filter cylinder for transporting sludge, and the operating parameters output from the trained model further include at least one of the following: a rotational speed of the first screw that is insufficient or overshooting its appropriate range, a rotational speed of the second screw that is insufficient or overshooting its appropriate range, a rotational torque of the first screw that is insufficient or overshooting its appropriate range, and a rotational torque of the second screw that is insufficient or overshooting its appropriate range.
In one aspect, the dewatering device is a belt press type dewatering device having a first filter cloth belt and a second filter cloth belt that sandwich and squeeze the sludge, and the operating parameters output from the trained model include at least one of the moving speeds of the first filter cloth belt and the second filter cloth belt, and the tensions of the first filter cloth belt and the second filter cloth belt.

In one aspect, a dehydration system for removing liquid from a suspension is provided, the dehydration system comprising: a suspension storage tank for storing a suspension; a flocculation device for stirring the suspension and a flocculant to form a flocculate; a dehydration device for removing liquid from the flocculate to form a turbid residue with a concentration of 3 to 50 wt %; an imaging device for generating image data of at least one of the suspension, the flocculate, the turbid residue, and the liquid removed from the suspension; and a control system having a trained model constructed by a machine learning algorithm, the control system comprising a storage device in which the trained model is stored, and a processing device for inputting the image data into the trained model and executing calculations for outputting from the trained model optimal operating parameters for the dehydration system that can bring the water content of the turbid residue discharged from the dehydration device within a target range.

In one aspect, a dehydration system for removing liquid from a suspension is provided, the dehydration system comprising: a suspension storage tank for storing a suspension; a flocculation device for stirring the suspension and a flocculant to form a flocculate; a concentration device for removing liquid from the flocculate to form a concentrate with a concentration of 3 to 15 wt %; a dehydration device for removing liquid from the concentrate to form a turbid residue with a concentration of 3 to 50 wt %; an imaging device for generating image data of at least one of the suspension, the flocculate, the concentrate, the turbid residue, and the liquid removed from the suspension; and a control system having a trained model constructed by a machine learning algorithm, the control system comprising: a storage device in which the trained model is stored; and a processing device for inputting the image data into the trained model and executing calculations for outputting from the trained model optimal operating parameters of the dehydration system that can bring the water content of the turbid residue discharged from the dehydration device within a target range.

In one embodiment, the concentrator is a rotating disc deliquator.
In one aspect, the control system is configured to input the image data and status data of the dewatering system into the trained model and cause the processing device to perform calculations to output from the trained model optimal operating parameters of the dewatering system that can keep the moisture content of the turbid residue discharged from the dewatering device within a target range.

In one aspect, a dehydration system for removing liquid from a suspension is provided, the dehydration system comprising: a flocculation device that mixes a suspension and a flocculant to form a floc, a dehydration device that removes liquid from the floc and forms a turbid residue, an imaging device that generates image data of the turbid residue discharged from the dehydration device, and a control system having a trained model constructed by a machine learning algorithm, the control system comprising a storage device that stores the trained model, and a processing device that inputs the image data into the trained model and executes calculations to output from the trained model an operating index value that indicates whether the operating state of the dehydration system is normal or abnormal.

In one aspect, the control system is configured to determine a running average of the driving metric value.
In one aspect, the control system determines that an anomaly is occurring in the operation of the dehydration system when a moving average of the operational index value exceeds a threshold value.
In one aspect, the control system resets the determination of an operational anomaly in the dehydration system when the moving average of the operational index value falls to a predetermined normal level.

According to the present invention, a trained model constructed by a machine learning algorithm generates optimal operating parameters that can keep the moisture content of the turbid residue (e.g., cake) remaining after removing liquid from a suspension (e.g., sludge) within a target range, and by applying the operating parameters to the flocculation device and/or dewatering device, the operation of the entire dewatering system can be optimized.

In particular, according to the present invention, instead of the operating parameters that were previously determined by an experienced operator based on visual inspection of the suspension, the dewatering system can use a trained model to generate optimal operating parameters that can bring the water content of the turbid residue within a target range. Therefore, the dewatering system can determine whether the coagulant addition rate is too high or too low, adjust the stirring speed for coagulation, adjust the operating parameters of the dewatering device, and so on in real time with the same or even greater accuracy as an experienced operator.

FIG. 1 illustrates one embodiment of a dewatering system. FIG. 1 is a schematic diagram illustrating an example of a trained model. FIG. 13 illustrates another embodiment of the dewatering system. FIG. 13 illustrates yet another embodiment of a dehydration system. 1 is a graph showing results of using a dehydration system. 1 is a graph showing results of using a dehydration system. 1 is a graph showing results of using a dehydration system. A graph showing the change over time in the operation index value output from the second trained model during operation of the dehydration system. 9 is a graph showing a change over time in a moving average of the driving index value shown in FIG. 8 .

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a diagram showing one embodiment of a dehydration system. The dehydration system includes a suspension storage tank 1 for storing a suspension, a flocculating device 3 for injecting a flocculant into the suspension and stirring the flocculant and the suspension to form a flocculate, a concentrating device 4 for removing liquid from the flocculate to form a concentrate, a dehydrating device 5 for further removing liquid from the concentrate made of the concentrated flocculate to form a turbid residue, and a control system 6 for controlling the operations of the flocculating device 3, the concentrating device 4, and the dehydrating device 5. The concentrating device 4 is a device for removing liquid from the flocculate formed by the flocculating device 3 to form a concentrate having a concentration of 3 to 15 wt %, and the dehydrating device 5 is a device for removing liquid from the flocculate formed by the flocculating device 3 or the concentrate formed by the concentrating device 4 to form a turbid residue having a concentration of 3 to 50 wt %.

The turbid residue is a low-water content material that remains after the liquid is removed from the suspension. The turbid residue that remains after the liquid is removed from the suspension by the dewatering device 5 is generally called cake. In the following explanation, the suspension is called sludge, and the turbid residue is called cake. Specific examples of sludge include sludge generated during the treatment of sewage, human waste, and industrial wastewater. Examples of suspensions other than sludge include industrial waste generated during the manufacture of industrial products such as food, cosmetics, and paper, or slurry.

In this embodiment, the concentrator 4 is a rotating disk type dewatering device (for example, a Slit Saver manufactured by Kendensha Co., Ltd.). This type of concentrator 4 is preferable because it is easy to grasp the changes in the state of the concentrated sludge, but the type of the concentrator 4 is not particularly limited. For example, the concentrator 4 may be a drum type, a screw type, or a belt type. The dewatering device 5 in this embodiment is composed of a screw press that pressurizes and dewaters the sludge. The screw press is an example of a pressurized dewatering machine that dewaters the sludge. The screw press can be of a pressurized type, a shaft sliding type, or a two-stage type. The dewatering device 5 shown in FIG. 1 is a shaft sliding type screw press. In addition to the screw press, other types such as a filter press and a centrifuge can also be used as the pressurized dewatering machine.

As shown in FIG. 1, the flocculation device 3 includes a flocculation mixing tank 7 that contains sludge and mixes the sludge with a flocculant, an agitator 8 for agitating the sludge in the flocculation mixing tank 7, a sludge inlet pipe (suspension inlet pipe) 14 connected to the flocculation mixing tank 7, and a pump 12 attached to the sludge inlet pipe 14. The agitator 8 includes an agitator blade 9 disposed in the flocculation mixing tank 7 and an agitator motor 10 connected to the agitator blade 9. The suspension storage tank 1 is connected to the flocculation mixing tank 7 by the sludge inlet pipe 14. The sludge is transferred from the suspension storage tank 1 to the flocculation mixing tank 7 through the sludge inlet pipe 14 by the pump 12. The flow rate of the sludge to the flocculation mixing tank 7 can be adjusted by operating the pump 12.

The pump 12 is connected to a control system 6, and the operation of the pump 12, i.e., the flow rate of sludge into the flocculation mixing tank 7, is controlled by the control system 6. A temperature sensor 22 is attached to the sludge inlet pipe 14. This temperature sensor 22 is configured to measure the temperature of the sludge introduced into the flocculation mixing tank 7. Furthermore, a sludge concentration meter (suspension concentration meter) 24 is attached to the sludge inlet pipe 14, and the concentration of suspended solids in the sludge introduced into the flocculation mixing tank 7 is measured by the sludge concentration meter 24.

The dewatering system includes a suspension imaging device 29 that generates image data of the sludge (suspension before treatment) in the suspension storage tank 1. The suspension imaging device 29 is a digital camera equipped with an image sensor (e.g., a CCD image sensor or a CMOS image sensor) that can generate still or continuous images of an object. Alternatively, the suspension imaging device 29 may be a hyperspectral camera that can image an object by breaking down light into wavelengths. The suspension imaging device 29 is connected to the control system 6, and the image data of the sludge (suspension) generated by the suspension imaging device 29 is sent to the control system 6.

The flocculation device 3 further includes a water supply line 26 connected to the flocculation mixing tank 7 and a flow control valve 27 attached to the water supply line 26. The water supply line 26 supplies water into the flocculation mixing tank 7 to dilute the sludge in the flocculation mixing tank 7. The flow rate of water supplied to the flocculation mixing tank 7 through the water supply line 26 is adjusted by the flow control valve 27. The flow control valve 27 is connected to the control system 6, and the operation of the flow control valve 27, i.e., the water flow rate, is controlled by the control system 6. In one embodiment, the water supply line 26 may be connected to the sludge introduction pipe 14, and water may be supplied directly to the sludge introduction pipe 14.

The flocculation device 3 further includes a flocculant supply device 28 connected to the flocculation mixing tank 7. The flocculant supply device 28 is configured to inject a flocculant into the sludge in the flocculation mixing tank 7 at a predetermined flow rate. The flow rate of the flocculant injected into the sludge in the flocculation mixing tank 7 is adjusted by the flocculant supply device 28. The flocculant supply device 28 is connected to the control system 6, and the operation of the flocculant supply device 28, i.e., the flow rate of the flocculant, is controlled by the control system 6. The sludge is agitated together with the flocculant by the agitator 8. By agitating the flocculant and sludge in the flocculation mixing tank 7, a flocculant is formed in which suspended solids in the sludge gather. The flocculant formed in the flocculation mixing tank 7 is generally called a flocculated floc.

The dewatering system includes an aggregate imaging device 31 that generates image data of the aggregates formed by the aggregation device 3. The aggregate imaging device 31 is a digital camera equipped with an image sensor (e.g., a CCD image sensor or a CMOS image sensor) that can generate still or continuous images of the object. Alternatively, the aggregate imaging device 31 may be a hyperspectral camera that can resolve light into wavelengths and image the object.

The agglomerate imaging device 31 faces the interior of the agglomeration mixing tank 7 and is arranged to generate image data of the agglomerate in the agglomeration mixing tank 7. The agglomerate imaging device 31 is connected to the control system 6, and image data of the agglomerate generated by the agglomerate imaging device 31 is sent to the control system 6. In one embodiment, the agglomerate imaging device 31 may face the outlet of the agglomeration mixing tank 7 and be arranged to generate image data of the agglomerate discharged from the agglomeration mixing tank 7.

The flocculation mixing tank 7 shown in FIG. 1 is a single-stage tank, but the flocculation mixing tank 7 may be a multi-stage tank. The agitation intensity of the sludge can be adjusted by the rotation speed of the agitator 8. The agitator 8 is connected to the control system 6, and the operation of the agitator 8, i.e., the agitation intensity of the sludge, is controlled by the control system 6. The flocculation mixing tank 7 may be provided with a side glass so that the state of the flocs inside can be observed.

The flocculation mixing tank 7 is connected to the thickener 4 by a sludge transfer pipe 18. The thickener 4 is disposed between the flocculation device 3 and the dewatering device 5. Sludge consisting of flocculants formed by the flocculation device 3 is transferred to the thickener 4 through the sludge transfer pipe 18. The flocculants are thickened and dewatered by the thickener 4. One example of the thickener 4 is a rotating disk type dewatering device.

The dehydration system further includes a concentrate imaging device 32 that generates image data of the concentrate (concentrated flocculate) formed by the concentrator 4. The concentrate imaging device 32 is a digital camera equipped with an image sensor (e.g., a CCD image sensor or a CMOS image sensor) that can generate still or continuous images of an object. Alternatively, the concentrate imaging device 32 may be a hyperspectral camera that can resolve light into wavelengths to capture an image of an object.

The concentrate imager 32 is disposed adjacent to the concentrator 4. More specifically, the concentrate imager 32 faces the inside of the concentrator 4 and is disposed to generate image data of the aggregate during the process of being concentrated by the concentrator 4. In one embodiment, the concentrate imager 32 may face the inlet of the concentrator 4 or the top of the concentrator 4 from the inlet of the concentrator 4 and be disposed to generate image data of the concentrate in a state in which liquid is being removed from the aggregate, or the concentrate imager 32 may face the outlet of the concentrator 4 and be disposed to generate image data of the aggregate after it has been concentrated by the concentrator 4, i.e., the concentrate. The concentrate imager 32 is connected to the control system 6, and the image data of the concentrate generated by the concentrate imager 32 is sent to the control system 6.

The dewatering system further includes a separated liquid imaging device 33 that generates image data of the liquid separated from the aggregate by the concentrator 4. The separated liquid imaging device 33 is a digital camera equipped with an image sensor (e.g., a CCD image sensor or a CMOS image sensor) that can generate still or continuous images of an object. Alternatively, the separated liquid imaging device 33 may be a hyperspectral camera that can image an object by decomposing light into wavelengths. The separated liquid imaging device 33 is disposed adjacent to the waste liquid drain 4a of the concentrator 4 and is disposed to generate image data of the liquid discharged from the concentrator 4. In one embodiment, the concentrator 4 has a liquid receiver (not shown) that receives the liquid discharged from the concentrator 4, and the separated liquid imaging device 33 may be disposed facing the inside of the liquid receiver. The separated liquid imaging device 33 is connected to the control system 6, and the image data of the liquid generated by the separated liquid imaging device 33 is sent to the control system 6.

The outlet of the concentrator 4 is located above the inlet 35 of the dehydrator 5, and the sludge consisting of the concentrate formed by the concentrator 4 is fed into the inlet 35 of the dehydrator 5. In this embodiment, the dehydrator 5 is a screw press. The dehydrator 5 as a screw press includes a filter cylinder 36, a screw shaft 37 arranged concentrically within the filter cylinder 36, a screw blade 38 fixed to the outer surface of the screw shaft 37, a screw motor 40 that rotates the screw shaft 37 and the screw blade 38 to send the sludge toward the discharge chamber 41, and a shaft sliding actuator 42 connected to the screw motor 40.

The filter cylinder 36 is made of a perforated plate such as punched metal. One end of the filter cylinder 36 is sealed by a blocking wall 44, and the other end of the filter cylinder 36 is connected to the discharge chamber 41. The inlet 35 is formed in the filter cylinder 36 and is adjacent to the blocking wall 44.

The screw shaft 37 extends through the filter tube 36. The screw shaft 37 has a truncated cone shape whose diameter gradually increases toward the downstream side. The screw shaft 37 extends through the blocking wall 44, and the end of the screw shaft 37 is connected to the screw motor 40. The screw motor 40 has a built-in inverter (not shown). The screw motor 40 is connected to the control system 6, and the operation of the screw motor 40, i.e., the rotational speed of the screw shaft 37 and the screw blades 38, is controlled by the control system 6.

The screw blade 38 is a single blade that extends in a spiral shape along the longitudinal direction of the screw shaft 37. A small gap is formed between the inner surface of the filter cylinder 36 and the screw blade 38, allowing the screw blade 38 to rotate without coming into contact with the filter cylinder 36. The sludge introduced into the filter cylinder 36 from the inlet 35 is transported through the filter cylinder 36 toward the discharge chamber 41 by the rotating screw blade 38.

The space through which the sludge is transported within the filter cylinder 36 is formed by the inner surface of the filter cylinder 36, the screw blades 38, and the screw shaft 37. The volume of this space gradually decreases along the direction in which the sludge moves. Therefore, as the sludge is moved through this space by the screw blades 38, it is squeezed and dewatered. The filtrate that passes through the filter cylinder 36 is collected by the filtrate receiver 45 located below the filter cylinder 36, and then discharged through the drain 46.

An annular back pressure plate 50 is disposed opposite the downstream end of the filter tube 36. This back pressure plate 50 has a truncated cone shape with a tapered surface for receiving the dewatered sludge transported inside the filter tube 36. A through hole through which the screw shaft 37 passes is formed in the center of the back pressure plate 50, and the back pressure plate 50 is disposed concentrically with the screw shaft 37. The back pressure plate 50 is not fixed to the screw shaft 37, and the back pressure plate 50 does not rotate.

The back pressure plate 50 is connected to a back pressure plate drive device 51. This back pressure plate drive device 51 is configured to move the back pressure plate 50 in the axial direction of the screw shaft 37. The gap between the back pressure plate 50 and the downstream end of the filter tube 36 is adjusted by the back pressure plate drive device 51. The back pressure plate drive device 51 is composed of, for example, a hydraulic cylinder or an electric cylinder. The back pressure plate drive device 51 is connected to the control system 6, and the operation of the back pressure plate drive device 51, i.e., the axial position of the back pressure plate 50, is controlled by the control system 6.

The shaft sliding actuator 42 is configured to move the screw motor 40 in the axial direction of the screw shaft 37. When the shaft sliding actuator 42 moves the screw motor 40 in the axial direction, the screw shaft 37 and the screw blades 38 connected to the screw motor 40 are moved in the axial direction of the screw shaft 37 within the filter tube 36.

Next, the operation of the dehydration device 5 will be described. Sludge consisting of the concentrate formed by the concentrator 4 is fed into the filter cylinder 36 from the inlet 35. The sludge is transported through the filter cylinder 36 toward the discharge chamber 41 by the rotating screw blades 38. As the sludge moves through the filter cylinder 36, it is squeezed and dehydrated. The filtrate that passes through the filter cylinder 36 is collected by the filtrate receiver 45 and discharged through the drain 46. The sludge is dehydrated in the filter cylinder 36 to form a cake. The cleaning device 55 supplies cleaning liquid to the outer circumferential surface of the filter cylinder 36 at preset time intervals.

The cake moving through the filter tube 36 is pressed against the back pressure plate 50. The cake is compressed by being prevented from moving by the back pressure plate 50. This compressed cake forms a plug 52 that seals the downstream end of the filter tube 36. The plug 52 applies back pressure to the following cake, thereby reducing the moisture content of the cake in the filter tube 36. While forming the plug 52 in the filter tube 36, the cake is pushed by the following cake and passes through the gap between the back pressure plate 50 and the downstream end of the filter tube 36, and is discharged little by little into the discharge chamber 41. The cake is discharged from the discharge chamber 41 through a discharge port 53 provided at the bottom of the discharge chamber 41. In this way, liquid is removed from the sludge, and a cake with a low moisture content is formed.

The gap between the back pressure plate 50 and the downstream end of the filter tube 36 changes depending on the axial position of the back pressure plate 50, and as a result, the compressive force applied to the sludge in the filter tube 36 changes. More specifically, when the gap between the back pressure plate 50 and the downstream end of the filter tube 36 becomes smaller, a larger force is required to push out the plug 52, and the compressive force applied to the sludge in the filter tube 36 increases. Therefore, the compressive force applied to the sludge can be adjusted not only by the rotation speed of the screw blade 38 but also by the position of the back pressure plate 50. A pressure sensor 61 is disposed in the filter tube 36, and the pressure of the sludge in the filter tube 36 is measured by the pressure sensor 61. The pressure sensor 61 is connected to the control system 6, and the measured value of the sludge pressure is sent to the control system 6. The position of the pressure sensor 61 is not particularly limited as long as it is in the filter tube 36, and may be in any of the front, middle, or rear stages in the filter tube 36, and multiple pressure sensors 61 may be disposed in the filter tube 36.

The plug 52 is formed from a cake with a low moisture content. If the moisture content of the cake that forms the plug 52 decreases, the plug 52 may harden and block the downstream end of the filter tube 36. In such a case, the shaft sliding actuator 42 can forcibly discharge the hard cake with a low moisture content by moving the screw shaft 37 and the screw blades 38 in the axial direction. As a result, the dewatering device 5 can operate stably and continuously.

The dehydration system includes a cake imaging device 64 that generates image data of the cake (turbid residue with low water content) formed by the dehydration device 5, and a dehydration filtrate imaging device 65 that generates image data of the filtrate discharged from the dehydration device 5. The cake imaging device 64 and the dehydration filtrate imaging device 65 are digital cameras equipped with an image sensor (e.g., a CCD image sensor or a CMOS image sensor) that can generate still images or continuous images of an object. Alternatively, the cake imaging device 64 and the dehydration filtrate imaging device 65 may be hyperspectral cameras that can resolve light into wavelengths and capture images of an object.

The cake imager 64 may be disposed adjacent to the discharge port 53 of the dewatering device 5, or adjacent to the back pressure plate 50 from which the cake is discharged. Furthermore, the cake imager 64 may be disposed in the discharge port 53 or the discharge chamber 41. The dewatered filtrate imager 65 is disposed adjacent to the drain 46 of the dewatering device 5. In one embodiment, the dewatered filtrate imager 65 may be disposed facing the inside of the filtrate receiver 45 and generate image data of the filtrate received in the filtrate receiver 45. Furthermore, in one embodiment, the dewatered filtrate imager 65 may be disposed facing the lower surface of the filter tube 36. The cake imager 64 and the dewatered filtrate imager 65 are connected to the control system 6, and the image data of the cake and the image data of the filtrate generated by the cake imager 64 and the dewatered filtrate imager 65 are sent to the control system 6.

The dehydration system further includes a filter tube imaging device 66 that generates image data of the exterior surface of the filter tube 36. The filter tube imaging device 66 is a digital camera equipped with an image sensor (e.g., a CCD image sensor or a CMOS image sensor) that can generate still or continuous images of an object. Alternatively, the filter tube imaging device 66 may be a hyperspectral camera that can resolve light into wavelengths to image an object.

The filter tube imaging device 66 is disposed above the filter tube 36 and generates image data of the outer surface of the filter tube 36 when the dewatering device 5 is squeezing the sludge. More specifically, the filter tube imaging device 66 is disposed to generate image data of the turbidity (solid matter in the sludge) leaking from the filter tube 36. In one embodiment, the filter tube imaging device 66 may be disposed to the side or below the filter tube 36. The filter tube imaging device 66 is connected to the control system 6, and the image data of the outer surface of the filter tube 36 generated by the filter tube imaging device 66 is sent to the control system 6.

The moisture content of the cake (turbid residue) formed in the dewatering device 5 varies depending on the state of the sludge introduced into the flocculation device 3 and the operating state of the dewatering system (i.e., the operating states of the flocculation device 3, the thickening device 4, and the dewatering device 5).

The control system 6 is configured to acquire image data generated by the above-mentioned imaging devices 29, 31, 32, 33, 64, 65, and 66, and optimize the operating parameters of the dehydration system based on this image data. The control system 6 includes a storage device 6a in which a trained model constructed by a machine learning algorithm is stored, and a processing device 6b that inputs image data into the trained model and executes calculations to output from the trained model the optimal operating parameters of the dehydration system that can keep the moisture content of the cake (turbid residue) discharged from the dehydration device 5 within a target range.

The storage device 6a includes a main storage device accessible to the processing device 6b, and an auxiliary storage device for storing programs, trained models, and data. The main storage device is, for example, a random access memory (RAM), and the auxiliary storage device is a storage device such as a hard disk drive (HDD) or a solid state drive (SSD). The processing device 6b is composed of a CPU (central processing unit) or a GPU (graphic processing unit), etc.

The control system 6 is composed of at least one computer. The at least one computer may be one server or multiple servers. The control system 6 may be an edge server connected to the imaging devices 29, 31, 32, 33, 64, 65, 66 by a communication line, a cloud server connected to the imaging devices 29, 31, 32, 33, 64, 65, 66 by a network such as the Internet, or a fog computing device (gateway, fog server, router, etc.) installed in a network connected to the imaging devices 29, 31, 32, 33, 64, 65, 66. The control system 6 may be multiple servers connected by a network such as the Internet. For example, the control system 6 may be a combination of an edge server and a cloud server.

The trained model is composed of a neural network. The storage device 6a stores a program for constructing the trained model according to a machine learning algorithm. The processing device 6b constructs the trained model by executing calculations according to instructions included in the program. Constructing the trained model according to the machine learning algorithm includes a process of optimizing parameters such as the weights of the neural network.

Examples of machine learning algorithms include the SVR method (support vector regression method), the PLS method (partial least squares method), the deep learning method, the random forest method, and the decision tree method, but the deep learning method is particularly suitable. The deep learning method is a learning method based on a neural network with multiple hidden layers. In this specification, machine learning using a neural network consisting of an input layer, two or more hidden layers, and an output layer is referred to as deep learning. By using the deep learning method, the state of sludge, flocculation, concentrated matter, etc., which was previously determined based on human eyes and experience, can be determined by a computer based on image data.

The trained model is a model constructed according to a machine learning algorithm using training data including multiple combinations of past image data generated by the imaging devices 29, 31, 32, 33, 64, 65, and 66 and optimal operating parameters of the dehydration system that can keep the moisture content of the cake (turbid residue) discharged from the dehydration device 5 within a target range. The past image data generated by the imaging devices are explanatory variables, and the optimal operating parameters of the dehydration system that can keep the moisture content of the cake (turbid residue) discharged from the dehydration device 5 within a target range are objective variables. The optimal operating parameters that constitute the training data are correct data, and are operating parameters determined by an experienced operator based on the state of sludge, flocculants, concentrates, etc. and the moisture content of the cake discharged from the dehydration device 5. The training data is also called training data or teacher data.

To optimize the parameters (weights, etc.) of the trained model, training data is prepared that includes many combinations of image data generated by the imaging devices 29, 31, 32, 33, 64, 65, 66 and optimal operating parameters that can bring the moisture content of the cake within a target range. The control system 6 uses this training data to construct a trained model through a machine learning algorithm. In addition to weights, the parameters of the trained model may include biases. The trained model constructed in this manner is stored in the storage device 6a.

The control system 6 acquires image data from the imaging devices 29, 31, 32, 33, 64, 65, and 66 at a predetermined cycle and stores the image data in the storage device 6a. The control system 6 inputs the image data into the trained model and outputs the optimal operating parameters from the trained model that can keep the moisture content of the cake within the target range.

In this embodiment, image data generated by all of the above-mentioned imaging devices 29, 31, 32, 33, 64, 65, and 66 is input to the trained model, but in one embodiment, image data generated by at least one of the imaging devices 29, 31, 32, 33, 64, 65, and 66 may be input to the trained model. In this case, image data of the same type as the image data input to the trained model is used to construct the trained model. For example, when image data generated by the aggregate imaging device 31 and the concentrate imaging device 32 is input to the trained model, the trained model is constructed using training data including past image data generated by the aggregate imaging device 31 and the concentrate imaging device 32.

Figure 2 is a schematic diagram showing an example of a trained model. The trained model is a neural network having an input layer 201, multiple hidden layers (also called intermediate layers) 202, and an output layer 203. The model shown in Figure 2 has four hidden layers 202, but the configuration of the trained model is not limited to the embodiment shown in Figure 2. The trained model may have five or more hidden layers 202.

Image data is input to the input layer 201 of the trained model. More specifically, the numerical values of each pixel constituting the image data are input to the input layer 201. If the image data is color image data, the numerical values representing the red, green, and blue colors of each pixel are input to the corresponding nodes (neurons) of the input layer 201 of the trained model. A control system 6 consisting of at least one computer performs calculations according to the algorithm of a multi-layer perceptron constituting a neural network, and the output layer 203 of the trained model outputs numerical values representing operating parameters that can keep the moisture content of the cake discharged from the dehydrator 5 within a target range. However, the configuration of the trained model shown in FIG. 2 is one example, and the present invention is not limited to the example shown in FIG. 2.

Specific examples of operating parameters output from the trained model are as follows:
- The flow rate of sludge to the coagulation mixing tank 7 is too high or too low relative to the appropriate range →Increase, maintain, or decrease the rotation speed of the pump 12. - The suitability of the coagulant is not appropriate → Maintain the use of the coagulant, or change the type of coagulant. - The flow rate of the coagulant is too high or too low relative to the appropriate range →Increase, maintain, or decrease the operating speed of the coagulant supply device 28. - The mixing speed of the agitator 8 is too high or too low relative to the appropriate range →Increase, maintain, or decrease the rotation speed of the agitator motor 10. - The flow rate of water supplied to the coagulation mixing tank 7 is too high or too low relative to the appropriate range →Increase, maintain, or decrease the opening of the flow control valve 27. - The operating speed of the thickener 4 is too high or too low relative to the appropriate range →Increase, maintain, or decrease. - The adjustment variable of the thickener 4 (e.g., compression plate pressure adjustment) is too high or too low relative to the appropriate range →Increase, maintain, or decrease. →Increase, maintain, or decrease. The rotation torque of the screw shaft 37 is insufficient or excessive relative to the appropriate range. →Increase, maintain, or decrease the current to the screw motor 40.

The above-mentioned driving parameters are examples, and the driving parameters output from the trained model may be only any of the above driving parameters.

Conventionally, a skilled operator would visually determine the state of the flocculants, concentrates, cakes, etc., and set the above operating parameters appropriately based on experience. In contrast, in this embodiment, the control system 6 inputs image data of the flocculants, concentrates, cakes, etc. into a model, and outputs optimal operating parameters from the trained model. By applying these operating parameters to the flocculation device 3, the concentration device 4, and the dehydration device 5, the operation of the entire dehydration system can be optimized.

In order to output more optimized operating parameters from the trained model, the training data used to construct the trained model may include, in addition to past image data, state data of the dehydration system when the past image data was generated. Specific examples of state data of the dehydration system are as follows.
- Temperature of the sludge (measured by temperature sensor 22)
Flow rate of sludge into the coagulation mixing tank 7 (estimated from the rotation speed of the pump 12)
- Sludge properties (pH, alkalinity, TS, VTS, SS, VSS, suspended solids)
・Physical properties of flocculant (cationicity, molecular weight)
Flow rate of flocculant (operating speed of flocculant supply device 28)
Stirring speed of the stirrer 8 (rotation speed of the stirring motor 10)
Flow rate of water to the flocculation mixing tank 7 (opening degree of the flow control valve 27)
Operation speed of the concentrator 4 (rotation speed of the drive motor of the concentrator 4)
Operating torque of the concentrator 4 (current value to the drive motor of the concentrator 4)
Flow rate of the liquid separated by the concentrator 4 (measured by a flow meter not shown)
Concentration of sludge thickened by the thickening device 4 (measured by a concentration sensor not shown)
Rotational speed of the screw shaft 37 Current value (torque value) to the screw motor 40
The pressure of the sludge in the filter cylinder 36 (measured by the pressure sensor 61)
Opening degree of the back pressure plate 50 (setting value of the back pressure plate driver 51)
Pressure of the back pressure plate 50 (measured by a pressure sensor not shown)
- Measurement value of cake moisture content (measured by a measuring device not shown or manually)

The above-mentioned items of status data of the dehydration system are examples, and the status data of the dehydration system may be only any of the above items.

The trained model is constructed according to a machine learning algorithm using training data including multiple combinations of past image data generated by at least one of the imaging devices 29, 31, 32, 33, 64, 65, and 66, state data of the dehydration system when the past image data was generated, and optimal operating parameters of the dehydration system that can keep the moisture content of the cake discharged from the dehydration device 5 within a target range. The past image data and the state data of the dehydration system when the past image data was generated are explanatory variables, and the optimal operating parameters of the dehydration system that can keep the moisture content of the cake discharged from the dehydration device 5 within a target range are objective variables. During operation of the dehydration system, the control system 6 inputs current image data generated by at least one of the imaging devices 29, 31, 32, 33, 64, 65, and 66 and the current state data of the dehydration system into the trained model, and causes the processing device 6b to execute a calculation for outputting the optimal operating parameters of the dehydration system from the trained model.

Figure 3 is a diagram showing another embodiment of the dehydration system. The dehydration system shown in Figure 3 differs from the embodiment in Figure 1, which uses a shaft-sliding screw press, in that a two-stage screw press is used as the dehydration device 5. In this embodiment, the shaft-sliding actuator 42 shown in Figure 1 is not provided. The configurations and operations of the embodiments not specifically described are the same as those of the embodiments shown in Figures 1 and 2, and therefore will not be described again.

The dewatering device 5, which is composed of a two-stage screw press, includes a filter cylinder 36, a first screw 71 and a second screw 72 that are arranged concentrically with the filter cylinder 36 within the filter cylinder 36 and transport sludge, a first screw motor 75 that rotates the first screw 71, and a second screw motor 76 that rotates the second screw 72 independently of the first screw 71. The operations of the first screw motor 75 and the second screw motor 76 are controlled by the control system 6.

The first screw motor 75 and the second screw motor 76 may be directly connected to the first screw 71 and the second screw 72, respectively, or may be connected to the first screw 71 and the second screw 72, respectively, via a torque transmission mechanism composed of a chain, sprockets, etc.

The filter cylinder 36 is formed from a screen (perforated plate) such as punched metal. An inlet 35 is formed at the upstream end of the filter cylinder 36. Sludge fed into the filter cylinder 36 from the inlet 35 is transported in a predetermined transport direction within the filter cylinder 36 by the rotating first screw 71 and second screw 72. The downstream end of the filter cylinder 36 is connected to the discharge chamber 41.

The second screw 72 is connected to the first screw 71 so as to be rotatable independently of the first screw 71. The first screw 71 and the second screw 72 extend through the filter tube 36 and the discharge chamber 41, respectively. The discharge chamber 41 is connected to the filter tube 36. A plug 77 made of cake is discharged from the filter tube 36 to this discharge chamber 41. The axial length of the second screw 72 is shorter than the axial length of the first screw 71. The first screw 71 has a first screw shaft 71A having a truncated cone shape whose diameter gradually increases along the sludge transport direction, and a first screw blade 71B fixed to the outer surface of the first screw shaft 71A. The second screw 72 has a cylindrical second screw shaft 72A and a second screw blade 72B fixed to the outer surface of the second screw shaft 72A.

The upstream end of the filter tube 36 is sealed by a blocking wall 44. The upstream end of the first screw shaft 71A extends through this blocking wall 44 and is connected to a first screw motor 75 for rotating the first screw 71. The second screw shaft 72A of the second screw 72 is arranged concentrically with the first screw shaft 71A. The outer diameter of the second screw shaft 72A is the same as the maximum diameter of the first screw shaft 71A. The downstream end of the second screw shaft 72A extends through the wall 41A that constitutes the discharge chamber 41 and is connected to a second screw motor 76 for rotating the second screw 72. The lower part of the discharge chamber 41 is connected to the discharge port 53.

The first screw motor 75 is connected to the control system 6. The first screw motor 75 has a built-in inverter (not shown), and the control system 6 is configured to be able to control the operation of the first screw motor 75 via the inverter. In other words, the control system 6 can control the rotation speed and rotation direction of the first screw motor 75 via the inverter. The control system 6 can issue a command to the first screw motor 75 to rotate the first screw 71 independently of the second screw 72.

The second screw motor 76 is also connected to the control system 6. The second screw motor 76 has a built-in inverter (not shown), and the control system 6 is configured to be able to control the operation of the second screw motor 76 via the inverter. In other words, the control system 6 can control the rotation speed and rotation direction of the second screw motor 76 via the inverter. The control system 6 can issue a command to the second screw motor 76 to rotate the second screw 72 independently of the first screw 71.

The first screw blade 71B extends helically along the axial direction of the first screw shaft 71A, and the second screw blade 72B extends helically along the axial direction of the second screw shaft 72A. The combined length of the portion of the first screw 71 to which the first screw blade 71B is fixed and the portion of the second screw 72 to which the second screw blade 72B is fixed is equal to or longer than the axial length of the filter tube 36.

A small gap is formed between the inner surface of the filter tube 36 and the first screw blade 71B, allowing the first screw blade 71B to rotate without contacting the filter tube 36. Similarly, a small gap is formed between the inner surface of the filter tube 36 and the second screw blade 72B, allowing the second screw blade 72B to rotate without contacting the filter tube 36. Sludge introduced into the filter tube 36 from the inlet 35 formed at the upstream end of the filter tube 36 can be transported toward the discharge chamber 41 by the rotating first screw blade 71B and second screw blade 72B.

The pitch of the second screw blade 72B is smaller than the pitch of the first screw blade 71B. Furthermore, the number of turns of the second screw blade 72B is less than three. In this embodiment, the winding direction (i.e., the helical direction) of the second screw blade 72B is opposite to the winding direction of the first screw blade 71B. Therefore, when the sludge introduced from the inlet 35 is sent to the discharge chamber 41, the second screw 72 is rotated in the opposite direction to the first screw 71, as shown in FIG. 3.

The winding direction of the second screw blade 72B may be the same as the winding direction of the first screw blade 71B. In this case, when the sludge introduced from the inlet 35 is sent to the discharge chamber 41, the second screw 72 is rotated in the same direction as the first screw 71.

The filter cylinder 36 is divided into a dewatering area 1A in which the first screw 71 is arranged, and a plug formation area 1B in which the second screw 72 is arranged. The space in which the sludge is transported in the dewatering area 1A is formed by the inner surface of the filter cylinder 36, the first screw blade 71B, and the first screw shaft 71A. The cross-sectional area of this transport space gradually decreases along the sludge transport direction, as shown in FIG. 3. Therefore, as the sludge introduced from the inlet 35 is transported through this transport space by the first screw blade 71B, the sludge is squeezed and dewatered. The filtrate that passes through the filter cylinder 36 is collected by the filtrate receiver 45 arranged below the filter cylinder 36. The filtrate receiver 45 is connected to a drain 46, and the filtrate collected by the filtrate receiver 45 is discharged from the dewatering device 5 via the drain 46.

The space through which the sludge is transported in the plug formation region 1B is formed by the inner surface of the filter cylinder 36, the second screw blade 72B, and the second screw shaft 72A. As shown in FIG. 3, the cross-sectional area of this transport space is constant. In the plug formation region 1B, plugs 77 are formed by the sludge (i.e., cake) dewatered in the dewatering region 1A. The plugs 77 apply back pressure to the subsequent cakes, thereby reducing the moisture content of the cakes in the filter cylinder 36. While forming plugs 77 in the plug formation region 1B, the cakes are pushed by the subsequent cakes and are discharged little by little into the discharge chamber 41. The cakes are discharged from the discharge chamber 41 through the discharge port 53 provided at the bottom of the discharge chamber 41. In this way, liquid is removed from the sludge, and a cake with a low moisture content is formed.

Similar to the embodiment shown in FIG. 1, the control system 6 is configured to input image data of sludge, flocculants, concentrates, etc. into a trained model and output optimal operating parameters that can keep the moisture content of the cake discharged from the dewatering device 5 within a target range.
Specific examples of operating parameters output from the trained model are as follows:
- The flow rate of sludge to the coagulation mixing tank 7 is too high or too low relative to the appropriate range →Increase, maintain, or decrease the rotation speed of the pump 12. - Suitability of the coagulant → Maintain the use of the coagulant, or change the type of coagulant. - The flow rate of the coagulant is too high or too low relative to the appropriate range →Increase, maintain, or decrease the operating speed of the coagulant supply device 28. - The mixing speed of the agitator 8 is too high or too low relative to the appropriate range →Increase, maintain, or decrease the rotation speed of the agitator motor 10. - The flow rate of water to be supplied to the coagulation mixing tank 7 is too high or too low relative to the appropriate range →Increase, maintain, or decrease the opening of the flow control valve 27. - The operating speed of the thickener 4 is too high or too low relative to the appropriate range →Increase, maintain, or decrease. - The adjustment variable of the thickener 4 (e.g., compression plate pressure adjustment) is too high or too low relative to the appropriate range →Increase, maintain, or decrease. →Increase, maintain, or decrease the rotation speed of the second screw motor 76. ·Whether the rotation torque of the first screw 71 is insufficient or excessive relative to the appropriate range. →Increase, maintain, or decrease the current to the first screw motor 75. ·Whether the rotation torque of the second screw 72 is insufficient or excessive relative to the appropriate range. →Increase, maintain, or decrease the current to the second screw motor 76.

The above-mentioned driving parameters are examples, and the driving parameters output from the trained model may be only any of the above driving parameters.

In order to output more optimized operating parameters from the model, the learning data used to construct the trained model may include, in addition to past image data, state data of the dehydration system when the past image data was generated. Specific examples of state data of the dehydration system are as follows:
- Temperature of the sludge (measured by temperature sensor 22)
Flow rate of sludge into the coagulation mixing tank 7 (rotation speed of pump 12)
- Sludge properties (pH, alkalinity, TS, VTS, SS, VSS, suspended solids)
・Physical properties of flocculant (cationicity, molecular weight)
Flow rate of flocculant (operating speed of flocculant supply device 28)
Stirring speed of the stirrer 8 (rotation speed of the stirring motor 10)
Flow rate of water to the flocculation mixing tank 7 (opening degree of the flow control valve 27)
- Operating speed of the concentrator 4 (rotation speed of the drive motor)
Operating torque of the concentrator 4 (current value to the drive motor)
Flow rate of the liquid separated by the concentrator 4 (measured by a flow meter not shown)
Concentration of sludge thickened by the thickening device 4 (measured by a concentration sensor not shown)
Rotational speed of the first screw 71 Current value (torque value) to the first screw motor 75
Rotational speed of the second screw 72 Current value (torque value) to the second screw motor 76
The pressure of the sludge in the filter cylinder 36 (measured by the pressure sensor 61)
- Measurement value of cake moisture content (measured by a measuring device not shown or manually)

The above-mentioned items of status data of the dehydration system are examples, and the status data of the dehydration system may be only any of the above items.

The trained model is constructed according to a machine learning algorithm using training data including multiple combinations of past image data generated by at least one of the imaging devices 29, 31, 32, 33, 64, 65, and 66, state data of the dehydration system when the past image data was generated, and optimal operating parameters of the dehydration system that can bring the moisture content of the cake discharged from the dehydration device 5 within a target range. During operation of the dehydration system, the control system 6 inputs current image data generated by at least one of the imaging devices 29, 31, 32, 33, 64, 65, and 66 and current state data of the dehydration system into the trained model, and causes the processing device 6b to execute a calculation for outputting optimal operating parameters of the dehydration system from the trained model.

In the embodiment shown in Figures 1 and 3, the dewatering system includes a thickener 4, but in one embodiment, the thickener 4 may be omitted. In this case, the flocs (flocs) formed in the flocculation device 3 are transported through the sludge transfer pipe 18 to the inlet 35 of the dewatering device 5.

Figure 4 is a diagram showing another embodiment of the dehydration system. In the dehydration system shown in Figure 4, a belt press type dehydration device is used as the dehydration device 5. The concentrating device 4 described above is not provided. The configuration and operation of the embodiments not specifically described are the same as those of the embodiments shown in Figures 1 and 2, so duplicated descriptions will be omitted.

The flocculation device 3 includes a granulation tank 80 that contains sludge, an agitator 8 for agitating the sludge in the granulation tank 80, a sludge inlet pipe 14 connected to the granulation tank 80, and a pump 12 attached to the sludge inlet pipe 14. The agitator 8 includes an agitator blade 9 disposed in the granulation tank 80 and an agitator motor 10 connected to the agitator blade 9. The suspension storage tank 1 is connected to the flocculation tank 80 by the sludge inlet pipe 14. The sludge is transferred from the suspension storage tank 1 to the granulation tank 80 through the sludge inlet pipe 14 by the pump 12. The flow rate of the sludge to the granulation tank 80 can be adjusted by operating the pump 12.

The pump 12 is connected to a control system 6, and the operation of the pump 12, i.e., the flow rate of sludge into the granulation tank 80, is controlled by the control system 6. A temperature sensor 22 is attached to the sludge inlet pipe 14. This temperature sensor 22 is configured to measure the temperature of the sludge introduced into the granulation tank 80. Furthermore, a sludge concentration meter (suspension concentration meter) 24 is attached to the sludge inlet pipe 14, and the concentration of suspended solids in the sludge introduced into the granulation tank 80 is measured by the sludge concentration meter 24.

The flocculation device 3 further includes a flocculant supply device 28 connected to the granulation tank 80. The flocculant supply device 28 is configured to inject a flocculant into the sludge in the granulation tank 80 at a predetermined flow rate. The flow rate of the flocculant injected into the sludge in the granulation tank 80 is adjusted by the flocculant supply device 28. The flocculant supply device 28 is connected to the control system 6, and the operation of the flocculant supply device 28, i.e., the flow rate of the flocculant, is controlled by the control system 6. The sludge is agitated together with the flocculant by the agitator blades 9 of the agitator 8. By agitating the flocculant and sludge in the granulation tank 80, a flocculant is formed in which suspended matter in the sludge gathers. The flocculant formed in the granulation tank 80 is generally called a pellet.

The dewatering system includes a suspension imaging device 29 that generates image data of the sludge (suspension before treatment) in the suspension storage tank 1. The suspension imaging device 29 is a digital camera equipped with an image sensor (e.g., a CCD image sensor or a CMOS image sensor) that can generate still or continuous images of an object. Alternatively, the suspension imaging device 29 may be a hyperspectral camera that can image an object by breaking down light into wavelengths. The suspension imaging device 29 is connected to the control system 6, and the image data of the sludge (suspension) generated by the suspension imaging device 29 is sent to the control system 6.

The dewatering system includes an agglomerate imaging device 31 that generates image data of the agglomerates (pellets) formed by the agglomeration device 3. The agglomerate imaging device 31 faces the inside of the granulation tank 80 and is arranged to generate image data of the agglomerates in the granulation tank 80. In one embodiment, the agglomerate imaging device 31 faces the outlet of the granulation tank 80 and may be arranged to generate image data of the agglomerates discharged from the granulation tank 80. The agglomerate imaging device 31 is connected to the control system 6, and the image data of the agglomerates generated by the agglomerate imaging device 31 is sent to the control system 6.

The agitation intensity of the sludge can be adjusted by the rotation speed of the agitator 8. The agitator 8 is connected to the control system 6, and the operation of the agitator 8, i.e., the agitation intensity of the sludge, is controlled by the control system 6.

The granulation tank 80 is connected to the belt press type dewatering device 5 by the sludge transfer pipe 18. The dewatering device 5 is equipped with an endless pretreatment filter cloth belt 82, a plurality of support rollers 83a to 83c that support the pretreatment filter cloth belt 82, an endless first filter cloth belt 86, first guide rollers 88a to 88e that support the first filter cloth belt 86, an endless second filter cloth belt 90, second guide rollers 91a to 91f that support the second filter cloth belt 90, a plurality of compression rollers 95A to 95E that support the first filter cloth belt 86 and the second filter cloth belt 90, and a drive device 98 that moves the first filter cloth belt 86 and the second filter cloth belt 90.

The pretreatment filter belt 82 is disposed upstream of the squeezing rollers 95A-95E in the sludge transport direction. A portion of the pretreatment filter belt 82 is supported by support rollers 83a and 83b so as to form an inclined portion 82A that is inclined upward along the sludge transport direction. The support roller 83a is connected to the drive shaft of the electric motor 100. When the electric motor 100 rotates the support roller 83a, the pretreatment filter belt 82 moves in a direction in which the sludge on the inclined portion 82A rises.

The drive device 98 includes an electric motor 98A, a drive wheel 98B fixed to the drive shaft of the electric motor 98A, and a belt 98C supported by the drive wheel 98B as a drive force transmission member. A chain may be used instead of the belt 98C. The belt 98C is connected to the first guide roller 88c and further to the second guide roller 91f. When the electric motor 98A rotates the drive wheel 98B, the rotation of the drive wheel 98B is transmitted to the first guide roller 88c and the second guide roller 91f by the belt 98C. The first guide roller 88c and the second guide roller 91f rotate at the same speed, so that the first filter cloth belt 86 and the second filter cloth belt 90 move in the same direction at the same speed.

The first filter cloth belt 86 is supported by first guide rollers 88a-88e and a number of compression rollers 95A-95E. The first guide roller 88d is connected to a first filter cloth tensioning device 101. This first filter cloth tensioning device 101 is configured to adjust the tension of the first filter cloth belt 86 by moving the position of the first guide roller 88d.

The second filter cloth belt 90 is supported by second guide rollers 91a-91f and a number of compression rollers 95A-95E. The second guide roller 91a is connected to a second filter cloth tensioning device 102. This second filter cloth tensioning device 102 is configured to adjust the tension of the second filter cloth belt 90 by moving the position of the second guide roller 91a.

The sludge consisting of the flocculants (pellets) formed by the flocculation device 3 is transferred through the sludge transfer pipe 18 onto the pretreatment filter cloth belt 82. As the sludge rises up the inclined portion 82A of the pretreatment filter cloth belt 82, the liquid in the sludge falls due to gravity. This process is the gravity dewatering process.

As the pretreatment filter cloth belt 82 moves, the sludge on the pretreatment filter cloth belt 82 is transferred onto the first filter cloth belt 86. The first filter cloth belt 86 and the second filter cloth belt 90, which extend between the multiple compression rollers 95A-95E, face each other. As the first filter cloth belt 86 moves, the sludge is sandwiched between the first filter cloth belt 86 and the second filter cloth belt 90. As the first filter cloth belt 86 and the second filter cloth belt 90 are wound around the compression rollers 95A-95E, the sludge is sandwiched between the first filter cloth belt 86 and the second filter cloth belt 90 and squeezed. As a result, the sludge becomes a cake (turbid residue) with a low moisture content. The cake with a low moisture content is discharged from the discharge outlet 53.

A filtrate receiver 45 is disposed below the compression rollers 95A-95E. The filtrate removed from the sludge is collected by the filtrate receiver 45. A drain 46 is connected to the filtrate receiver 45, and the filtrate collected by the filtrate receiver 45 is discharged from the dehydrator 5 via the drain 46.

The dewatering system includes a cake imaging device 64 that generates image data of the cake (turbid residue with low water content) formed by the dewatering device 5, and a dewatered filtrate imaging device 65 that generates image data of the filtrate discharged from the dewatering device 5. The cake imaging device 64 is disposed adjacent to the discharge outlet 53 of the dewatering device 5, and the dewatered filtrate imaging device 65 is disposed adjacent to the drain 46 of the dewatering device 5. The cake imaging device 64 may be disposed within the discharge outlet 53. The cake imaging device 64 and the dewatered filtrate imaging device 65 are connected to the control system 6, and the image data of the cake and the image data of the filtrate generated by the cake imaging device 64 and the dewatered filtrate imaging device 65 are sent to the control system 6.

The control system 6 is configured to acquire image data generated by the imaging devices 29, 31, 64, 65 from the imaging devices 29, 31, 64, 65, and optimize the operation parameters of the dehydration system based on this image data. The control system 6 includes a storage device 6a in which a trained model constructed by a machine learning algorithm is stored, and a processing device 6b that inputs image data into the trained model and executes calculations to output from the trained model the optimal operation parameters of the dehydration system that can keep the moisture content of the cake (turbid residue) discharged from the dehydration device 5 within a target range.

The trained model is a model constructed according to a machine learning algorithm using training data including multiple combinations of past image data generated by the imaging devices 29, 31, 64, and 65 and optimal operating parameters of the dehydration system that can keep the moisture content of the cake (turbid residue) discharged from the dehydration device 5 within a target range. The past image data generated by the imaging devices 29, 31, 64, and 65 are explanatory variables, and the optimal operating parameters of the dehydration system that can keep the moisture content of the cake (turbid residue) discharged from the dehydration device 5 within a target range are objective variables. The optimal operating parameters that make up the training data are correct data, and are operating parameters determined by an experienced operator based on the state of sludge, flocculants, etc., and the moisture content of the cake discharged from the dehydration device 5.

To optimize the parameters (weights, etc.) of the trained model, training data is prepared that includes many combinations of image data generated by the imaging devices 29, 31, 64, 65 and optimal operating parameters that can bring the moisture content of the cake within a target range. The control system 6 uses this training data to construct a trained model through a machine learning algorithm. In addition to weights, the parameters of the trained model may include biases. The trained model constructed in this manner is stored in the storage device 6a.

The control system 6 acquires image data from the imaging devices 29, 31, 64, and 65 at a predetermined cycle and stores the image data in the storage device 6a. The control system 6 inputs the image data into the trained model and outputs the optimal operating parameters from the trained model that can keep the moisture content of the cake within the target range.

In this embodiment, image data generated by all of the above-mentioned imaging devices 29, 31, 64, and 65 is input to the trained model, but in one embodiment, image data generated by at least one of the imaging devices 29, 31, 64, and 65 may be input to the trained model. In this case, image data of the same type as the image data input to the trained model is used to construct the trained model. For example, when image data generated by the aggregate imaging device 31 is input to the trained model, the trained model is constructed using training data including past image data generated by the aggregate imaging device 31.

Similar to the embodiment shown in FIG. 1, the control system 6 is configured to input image data into a trained model and output optimal operating parameters from the trained model that can keep the moisture content of the cake discharged from the dewatering device 5 within a target range.
Specific examples of operating parameters output from the trained model are as follows:
- The flow rate of sludge into the granulation tank 80 is insufficient or excessive relative to the appropriate range → Increase, maintain, or decrease the rotation speed of the pump 12. - Suitability of the coagulant → Maintain the use of the coagulant, or change the type of coagulant. - The flow rate of the coagulant is insufficient or excessive relative to the appropriate range → Increase, maintain, or decrease the operating speed of the coagulant supply device 28. - The stirring speed of the agitator 8 is insufficient or excessive relative to the appropriate range → Increase, maintain, or decrease the rotation speed of the agitator motor 10. - The moving speed of the first filter cloth belt 86 and the second filter cloth belt 90 → Increase, maintain, or decrease the rotation speed of the motor 98A. - The tension of the first filter cloth belt 86 and the second filter cloth belt 90 → Change or maintain the positions of the guide rollers 88d, 91a.

The above-mentioned driving parameters are examples, and the driving parameters output from the trained model may be only any of the above driving parameters.

In order to output more optimized operating parameters from the model, the learning data used to build the model may include, in addition to past image data, state data of the dehydration system when the past image data was generated. Specific examples of state data of the dehydration system are as follows.
- Temperature of the sludge (measured by temperature sensor 22)
Flow rate of sludge into the granulation tank 80 (rotation speed of pump 12)
- Sludge properties (pH, alkalinity, TS, VTS, SS, VSS, suspended solids)
・Physical properties of flocculant (cationicity, molecular weight)
Flow rate of flocculant (operating speed of flocculant supply device 28)
Stirring speed of the stirrer 8 (rotation speed of the stirring motor 10)
The moving speed of the first filter cloth belt 86 and the second filter cloth belt 90 (the rotation speed of the motor 98A)
Positions of guide rollers 88d and 91a

The above-mentioned items of status data of the dehydration system are examples, and the status data of the dehydration system may be only any of the above items.

The trained model is constructed according to a machine learning algorithm using training data including multiple combinations of past image data generated by at least one of the imaging devices 29, 31, 64, and 65, state data of the dehydration system when the past image data was generated, and optimal operating parameters of the dehydration system that can bring the moisture content of the cake discharged from the dehydration device 5 within a target range. During operation of the dehydration system, the control system 6 inputs current image data generated by at least one of the imaging devices 29, 31, 64, and 65 and current state data of the dehydration system into the trained model, and causes the processing device 6b to execute a calculation to output the optimal operating parameters of the dehydration system from the model.

When sludge is coagulated with a coagulant such as a polymer to form coagulated flocs, and the sludge containing the coagulated flocs is subjected to solid-liquid separation, the solid-liquid separation performance differs depending on the type and properties of the sludge, the coagulant used, the type of solid-liquid separation device, etc. If a change occurs in the solid-liquid separation state, the plant operator on-site must quickly detect the change and adjust and change the operating conditions as appropriate, thereby maintaining the optimized state of solid-liquid separation as continuously as possible. When judging and evaluating this "solid-liquid separation state of sludge," the points where the change in state is most pronounced and noticeable vary greatly depending on the type of sludge, so the examples were carried out at dewatering treatment facilities that handle the following three different types of sludge. It goes without saying that the items adopted and imported as input signals in these three examples may be increased or decreased due to seasonal fluctuations in sludge properties or changes in the operating environment, thereby improving efficiency.

In the above Examples 1 to 3, each skilled operator who operates and manages the shaft sliding screw press shown in Figure 1, which is operated in a dewatering treatment facility for three different types of sludge, optimized the operating conditions while referring to the above-mentioned operating parameters and the image data (seven locations) for a certain period of time. Each time the operating conditions were changed for optimization, the operating parameters and image data on which the operating conditions were changed were recorded, and multivariate analysis was performed based on the statistical data of these records. Furthermore, the contribution of each operating parameter or each image data that was the basis for changing the operating operation items for operation optimization was statistically organized, and operating parameters or image data that were particularly important and necessary as input data for adoption in each of the three sludge dewatering treatment facilities were extracted. As a result, it was found that the input data shown below for each treatment facility was almost necessary and sufficient data for operation optimization. When these data were imported as input data for this dewatering system, the probability that the operating parameter contents output as output data matched the actual operation contents of the skilled operator was 93% or more for all three treatment facilities. The operating conditions of Examples 1 to 3 are explained individually below.

From the operation record data, the output data from the dehydration system, and the statistical analysis results, it was found that the solid-liquid separation state of the sewage mixed raw sludge in Example 1 can be roughly judged and evaluated by three elements: "pressure inside the dehydrator,""screw shaft torque," and "image of filtrate discharged near the screw outlet." The matching rate between the output data of the dehydration system using these three input data and the actual operation contents of the skilled operator reached 94%. Figure 5 shows the test results of Example 1. In general, mixed raw sludge has a relatively large amount of 10 to 40% of coarse floating matter of 74 μm or more (the larger the better the solid-liquid separation), which contributes to solid-liquid separation, and forms a firm dehydrated cake with a firm texture, and the cake moisture content is dehydrated to about 60 to 70%. For this reason, it is considered that the dehydration performance is strongly reflected in the pressure inside the dehydrator and the screw shaft torque, and the dehydrated filtrate image (later stage) characteristically shows the solid-liquid separation state (the dripping of the filtrate) at the final stage of the dehydration process.
The moisture content of the dehydrated cake during the operation period of this test achieved the target moisture content for a given sludge flow rate.

From the operation record data, the output data from this dehydration system, and the results of their statistical analysis, it was found that in the mixed biomass methane fermentation sludge of Example 2, the solid-liquid separation state can be roughly determined and evaluated using only the "image of the upper part of the thickening device." The matching rate between the output data of this dehydration system using only this input data and the actual operation content of an experienced operator reached 93%. Figure 6 shows the test results for Example 2.

In general, mixed biomass methane fermentation sludge contains a large amount of processed food products and food residues with relatively high salt concentrations, so the appropriate flocculant addition rate may exceed 4% per solid. Furthermore, the organic matter fed into the methane fermentation tank varies greatly in both quality and quantity every day, so controlling the flocculant injection rate within an appropriate range is the most important factor for the solid-liquid separability of the sludge. It is believed that the movement of the flocculant-concentrated sludge in the draining zone of the thickener (spreading, rolling, solid-liquid interface, etc.) was able to accurately express the fluctuation in the appropriate injection rate in real time. It is believed that an experienced operator could quickly detect changes in solid-liquid separability caused mainly by fluctuations in the appropriate flocculant injection rate by constantly monitoring images from the top of the draining zone of the thickener, and re-optimize the operating conditions of the dewatering system.
The moisture content of the dehydrated cake during the operation period of this test achieved the target moisture content for a given sludge flow rate.

From the operation record data, the output data from the dehydration system, and the statistical analysis results, it was found that the solid-liquid separation state of the excess sludge in the activated sludge treatment facility for wastewater from a drinking water manufacturing plant in Example 3 can be roughly judged and evaluated by comprehensively evaluating six elements: "image of the flocculated flocs in the flocculation mixing tank,""image of the top of the thickener,""state of turbidity leakage from the dehydrator,""image showing the turbidity of the dehydration filtrate,""pH of the sludge," and "pressure inside the dehydrator." The matching rate between the output data of the dehydration system using these six input data and the actual operation contents of an experienced operator reached 95%. The test results of Example 3 are shown in Figure 7. The excess sludge from the drinking water manufacturing plant wastewater treatment facility that was the subject of this study had an extremely low ratio of coarse suspended matter of 74 μm or more, which contributes to solid-liquid separation, at less than 3%, and a relatively high VTS of 88%, which also affects solid-liquid separation (the smaller the VTS, the better the solid-liquid separation), making the sludge difficult to dewater. However, by using this dewatering system to output optimal operating parameters based on data mainly consisting of images of the shape and texture of the coagulated flocs, the movement of the sludge at the top of the thickening device, and the state of SS leakage from the dewatering device, which are known from experience by experienced operators, it is believed that it is possible to operate the sludge at the same level as an experienced operator who can understand the characteristics of this difficult-to-dewaterable sludge.
The moisture content of the dehydrated cake during the operation period of this test achieved the target moisture content for a given sludge flow rate.

The above-described embodiment outputs the current optimal operating parameters while achieving the target moisture content, but it can also be applied to predict future operating conditions. For example, it is possible to predict the moisture content of future dehydrated cake discharged from this dehydration system and possible future malfunctions of equipment in the dehydration process.

In one embodiment, the control system 6 may include a trained model for determining whether the operation of the dehydration system is normal or abnormal, in addition to or instead of the trained model described above that outputs optimal operating parameters for the dehydration system. If an abnormality occurs in the operating state of the dehydration system, the moisture content of the resulting dehydrated cake may increase, which may have a significant impact on downstream processing. Although such operational abnormalities do not occur frequently, it is important to correctly detect any operational abnormality in the dehydration system when it does occur.

Therefore, this embodiment provides a dehydration system that can correctly detect operational abnormalities in the dehydration system using a trained model. In the following description, the trained model that outputs optimal operating parameters for the dehydration system is referred to as the first trained model, and the trained model for determining whether the operation of the dehydration system is normal or abnormal is referred to as the second trained model.

The second trained model is stored in the storage device 6a of the control system 6. In one example, the second trained model is composed of a neural network. The storage device 6a stores a program for constructing the second trained model according to a machine learning algorithm. The processing device 6b constructs the second trained model by executing calculations according to instructions included in the program. The details of the second trained model are similar to those of the first trained model, so a duplicated description will be omitted.

The second trained model is a model constructed according to a machine learning algorithm using training data including multiple combinations of past image data generated by the cake imaging device 64 and operation index values indicating that the operating state of the dewatering system is normal or abnormal. As shown in Figures 1, 3, and 4, the cake imaging device 64 is arranged to generate image data of the cake (turbid residue with low moisture content) formed by the dewatering device 5. In one example, the cake imaging device 64 may be arranged above a discharge conveyor (not shown) for conveying the cake formed by the dewatering device 5, and arranged to generate image data of the cake on the discharge conveyor.

The past image data generated by the cake imaging device 64 is an explanatory variable, and the operation index value indicating whether the operating state of the dehydration system is normal or abnormal is the objective variable. Whether the operating state of the dehydration system is normal or abnormal is determined by the operator. Examples of the operation index value include a first numerical value (e.g., 0) indicating that the operating state of the dehydration system is normal, and a second numerical value (e.g., 1) indicating that the operating state of the dehydration system is abnormal. Causes of an abnormal operating state of the dehydration system include poor dissolution concentration of the coagulant, unexpected fluctuations in sludge concentration, and clogging of sludge in the dehydration device 5.

In order to optimize the parameters (weights, etc.) of the second learned model, learning data (also called training data or teacher data) is prepared, which includes a large number of combinations of image data generated by the cake imaging device 64 and operation index values that indicate whether the operating state of the dehydration system is normal or abnormal. The control system 6 uses this learning data to construct the second learned model by a machine learning algorithm. The parameters of the second learned model may include bias in addition to weights. The second learned model constructed in this manner is stored in the storage device 6a.

When the dehydration system is operating, the control system 6 acquires current image data from the cake imaging device 64 at a predetermined cycle and stores it in the memory device 6a. The control system 6 inputs the image data into the second trained model and causes the processing device 6b to execute a calculation to output an operation index value from the second trained model indicating whether the operating status of the dehydration system is normal or abnormal.

Figure 8 is a graph showing the change over time in the operation index value output from the second trained model while the dehydration system is operating. In Figure 8, the vertical axis represents the operation index value, and the horizontal axis represents the operation time. The interval at which the judgment result is output by the second trained model is set as appropriate. Figure 8 shows an example in seconds, but the interval may be longer, such as in minutes, hours, or days. In the first half of the graph in Figure 8, the value 0 continues, indicating that the operation is normal, but gradually the value 1 is output, indicating an abnormal operation. As shown in Figure 8, while the same value is continuously output from the second trained model, different values may sometimes be output. This is because the second trained model was unable to correctly judge the operating state of the dehydration system.

Figure 9 is a graph showing the change over time in the moving average of the operation index value shown in Figure 8. In Figure 8, the vertical axis represents the moving average of the operation index value, and the horizontal axis represents the operation time. The control system 6 calculates the moving average of the operation index value output from the second trained model. The time interval of the moving average is set appropriately. The control system 6 compares the moving average of the operation index value with a threshold value, and when the moving average of the operation index value exceeds the threshold value, it determines that an abnormality has occurred in the operation of the dehydration system. The control system 6 then issues an alarm indicating an abnormality in the operation of the dehydration system. The alarm allows the operator to know that an abnormality has occurred in the operation of the dehydration system, and the abnormality can be resolved by taking appropriate measures for the treatment system.

As can be seen from FIG. 9, the moving average of the operation index value continues to fluctuate even after it exceeds the threshold value, so it exceeds and falls below the threshold value. Therefore, once the control system 6 determines that an abnormality has occurred in the operation of the dehydration system, it does not perform an operation abnormality determination operation and does not cancel the alarm unless the moving average of the operation index value falls to a predetermined normal level. When the operation abnormality of the dehydration system is resolved, the moving average of the operation index value gradually falls. When the moving average of the operation index value falls to the predetermined normal level, the control system 6 resets the operation of determining an operation abnormality of the dehydration system and resumes the operation of determining an operation abnormality. Furthermore, the control system 6 cancels the alarm.

According to this embodiment, the control system 6 can correctly determine that an abnormality has occurred in the dehydration system based on the operation index value output from the second trained model.

The control system 6 may include either or both of a first trained model that outputs optimal operating parameters for the dehydration system and a second trained model that determines whether the operation of the dehydration system is normal or abnormal.

The above-described embodiments have been described with the aim of enabling a person having ordinary knowledge in the technical field to which the present invention pertains to practice the present invention. Various modifications of the above-described embodiments would naturally be possible for a person skilled in the art, and the technical concept of the present invention may also be applied to other embodiments. Therefore, the present invention is not limited to the described embodiments, but is to be interpreted in the broadest scope in accordance with the technical concept defined by the scope of the claims.

LIST OF SYMBOLS 1 Suspension storage tank 3 Flocculation device 4 Concentration device 5 Dewatering device 6 Control system 6a Storage device 6b Treatment device 7 Flocculation mixing tank 8 Agitator 9 Agitation blade 10 Agitation motor 12 Pump 14 Sludge inlet pipe 18 Sludge transport pipe 22 Temperature sensor 24 Sludge concentration meter 26 Water supply line 27 Flow control valve 28 Flocculant supply device 29 Suspension imaging device 31 Flocculation imaging device 32 Concentrate imaging device 33 Separation liquid imaging device 35 Inlet 36 Filter cylinder 37 Screw shaft 38 Screw blade 40 Screw motor 41 Discharge chamber 42 Shaft sliding actuator 44 Blocking wall 45 Filtrate receiver 46 Drain 50 Back pressure plate 51 Back pressure plate drive device 52 Plug 53 Discharge port 55 Cleaning device 61 Pressure sensor 64 Cake imaging device 65 Dewatered filtrate imaging device 66 Filter cylinder imaging device 71 First screw 71A First screw shaft 71B First screw blade 72 Second screw 72A Second screw shaft 72B Second screw blade 75 First screw motor 76 Second screw motor 77 Plug 80 Granulation tank 82 Pretreatment filter cloth belts 83a to 83c Support roller 86 First filter cloth belts 88a to 88e First guide roller 90 Second filter cloth belts 91a to 91f Second guide rollers 95A to 95E Squeezing roller 98 Drive device 98A Electric motor 98B Drive wheel 98C Belt 100 Electric motor 101 First filter cloth tensioning device 102 Second filter cloth tensioning device

Claims (8)

1. A dewatering system for removing liquid from sludge, comprising:
a suspension storage tank for storing sludge;
a flocculation device for mixing the sludge and a flocculant to form a floc;
a dewatering device for removing liquid from the agglomerate to form a cake having a concentration of 3 to 50 wt %;
an imaging device for generating image data of the liquid removed from the sludge;
A control system having a trained model constructed by a machine learning algorithm,
The control system includes a storage device in which the trained model is stored, and a processing device that inputs the image data and status data into the trained model and executes a calculation to output, from the trained model, optimal operating parameters of the dehydration system that can keep the moisture content of the cake discharged from the dehydration device within a target range;
The status data includes a flow rate of the sludge to the flocculation device, a property of the sludge, and a flow rate of the flocculant. 1. A dewatering system for removing liquid from sludge, comprising:
a suspension storage tank for storing sludge;
a flocculation device for mixing the sludge and a flocculant to form a floc;
a concentrator for removing liquid from the agglomerate to form a concentrate having a concentration of 3-15 wt %;
a dewatering device for removing liquid from the concentrate to form a cake having a concentration of 3 to 50 wt %;
an imaging device for generating image data of the liquid removed from the sludge;
A control system having a trained model constructed by a machine learning algorithm,
The control system includes a storage device in which the trained model is stored, and a processing device that inputs the image data and status data into the trained model and executes a calculation to output, from the trained model, optimal operating parameters of the dehydration system that can keep the moisture content of the cake discharged from the dehydration device within a target range;
The status data includes a flow rate of the sludge to the flocculation device, a property of the sludge, and a flow rate of the flocculant. The dehydration system according to claim 1 or 2, wherein the image data of the liquid removed from the sludge is image data of the filtrate discharged from the dehydration device. The dehydration system according to claim 3, wherein the dehydration device is a screw press or a belt press type dehydration device. The operating parameters output from the trained model are
The flow rate of sludge to the flocculation device is in excess of or insufficient to the proper range,
suitability of the flocculant;
The flow rate of the flocculant is insufficient or excessive with respect to the appropriate range,
The stirring speed of the flocculation device is in excess of or insufficient to the appropriate range,
The flow rate of the water supplied to the flocculation device is in excess of or insufficient to the appropriate range.
The dehydration system of claim 1 , comprising at least one of: The dehydration device is a screw press including a screw shaft disposed in a filter cylinder and a screw blade fixed to an outer surface of the screw shaft,
The operating parameters output from the trained model are
The rotation speed of the screw shaft is in excess of or insufficient to the appropriate range,
The rotation torque of the screw shaft is insufficient or excessive with respect to the appropriate range.
The dehydration system of claim 1 , further comprising at least one of: The dewatering device is a two-stage screw press including a first screw and a second screw arranged in a filter cylinder for transporting sludge,
The operating parameters output from the trained model are
The rotation speed of the first screw is insufficient or excessive with respect to the appropriate range,
The rotation speed of the second screw is insufficient or excessive with respect to the appropriate range.
the rotational torque of the first screw being insufficient or excessive relative to an appropriate range;
the rotation torque of the second screw being insufficient or excessive relative to the appropriate range;
The dehydration system of claim 1 , further comprising at least one of: The dewatering device is a belt press type dewatering device having a first filter cloth belt and a second filter cloth belt that sandwich and squeeze the sludge,
The operating parameters output from the trained model are
the moving speeds of the first filter cloth belt and the second filter cloth belt;
tension of the first filter cloth belt and the second filter cloth belt;
The dehydration system of claim 1 , comprising at least one of:

Images