JP7782196B2 - Gas concentration estimation device, gas concentration estimation method, and program - Google Patents

Description

The present invention relates to a gas concentration estimation device, a gas concentration estimation method, and a program.

Generally, in order to achieve both energy conservation and pollution prevention in a combustion furnace, it is important to control the gas concentrations (CO concentration and _O2 concentration) inside the combustion furnace, and for this purpose, gas concentrations are measured using various methods.

For example, a method of measuring gas concentration using a laser gas analyzer is known (see, for example, Non-Patent Document 1). While laser gas analyzers can measure gas concentration in real time, they have problems such as being expensive and data loss occurring when the laser receiving section is blocked.

Methods of measuring gas concentration using, for example, an infrared gas analyzer are also known. While infrared gas analyzers are less expensive than laser gas analyzers, they measure the gas concentration of gas discharged into the flue, and therefore can only measure gas concentrations at times delayed from the actual gas concentration inside the combustion furnace. Specifically, if the gas concentration inside the combustion furnace at time t is y(t), then the infrared gas analyzer measures the gas concentration y(t-τ) a certain time lag τ from the current time t.

Yoshitaka Yuki, Akihiro Murata, "Optimal Combustion Control Solution Using the TDLS200 Laser Gas Analyzer," Yokogawa Technical Report 53(1), 19-22, 2010

Infrared gas analyzers are inexpensive and easy to introduce, but as mentioned above, there is a time delay in measuring gas concentrations. For this reason, controlling a combustion furnace based on gas concentrations measured by an infrared gas analyzer may result in a deterioration in performance in terms of combustion efficiency and energy conservation.

One embodiment of the present invention has been made in consideration of the above points, and aims to estimate gas concentrations without time delay from gas concentrations measured by an infrared gas analyzer.

To achieve the above-mentioned objective, one embodiment of a gas concentration estimation device is a gas concentration estimation device that estimates gas concentrations for controlling a combustion furnace, and includes: a receiving means that receives a time-delayed gas concentration measurement value obtained by measuring the concentration of gas discharged from the combustion furnace into the flue using an infrared gas analyzer; and one or more physical quantity measurement values obtained by measuring one or more predetermined physical quantities related to the combustion furnace using sensors; and an estimation means that uses a model for estimating gas concentrations without a time delay to calculate an estimate of the time-delayed gas concentration measurement value as a gas concentration for controlling the combustion furnace from the time-delayed gas concentration measurement value and the one or more physical quantity measurement values.

Gas concentrations can be estimated without time delay from the gas concentrations measured by an infrared gas analyzer.

1 is a diagram showing an example of the overall configuration of a combustion control system according to an embodiment of the present invention; 3A and 3B are diagrams schematically showing the relationship between the excess air ratio and heat loss and the relationship between the excess air ratio and thermal efficiency. FIG. 1 is a diagram illustrating an example of a hardware configuration of a gas concentration estimation device according to an embodiment of the present invention. FIG. 2 is a diagram illustrating an example of a functional configuration of the gas concentration estimation device according to the present embodiment. 10 is a flowchart illustrating an example of a model estimation process according to the present embodiment. 4 is a flowchart illustrating an example of a gas concentration estimation process according to the present embodiment.

One embodiment of the present invention will be described below. The following describes a combustion control system 1 that includes a gas concentration estimation device 10 that estimates the gas concentration in a combustion furnace at the current time from the gas concentration (i.e., gas concentration including a time delay) measured using an infrared gas analyzer when the gas concentration of gas (exhaust gas) discharged from a combustion furnace into the flue is measured.

<Overall configuration of combustion control system 1>
The overall configuration of a combustion control system 1 according to this embodiment is shown in Fig. 1. As shown in Fig. 1, the combustion control system 1 according to this embodiment includes a gas concentration estimation device 10, an exhaust gas concentration measuring instrument 20, a control device 30, a valve 40, a peripheral sensor 50, a combustion furnace 60, and a flue 70.

The combustion furnace 60 may be, for example, a heating furnace, boiler, incineration, or other type of equipment in a plant. The flue 70 is a facility that serves as a passageway for gas (exhaust gas) emitted from the combustion furnace 60. The exhaust gas that passes through the flue 70 is drawn into the exhaust gas concentration measuring instrument 20 and is ultimately emitted into equipment such as a chimney.

The flue gas concentration measuring instrument 20 is an infrared gas analyzer that measures the gas concentration (e.g., CO concentration or _O2 concentration) of the flue gas in the flue and outputs the measurement value to the gas concentration estimation device 10. Here, since the flue gas concentration measuring instrument 20 measures the gas concentration of the flue gas drawn in from the flue 70, a time delay occurs in the measurement value. Specifically, if the gas concentration in the combustion furnace 60 at time t is y(t), the flue gas concentration measuring instrument 20 measures the gas concentration y(t-τ) in the combustion furnace 60 a certain time lag τ ago from the current time t.

The time delay τ is generally the time it takes for the flue gas discharged from the combustion furnace 60 to reach the flue gas concentration measuring instrument 20, but the time required for the flue gas concentration measuring instrument 20 to make a measurement may also be taken into consideration. That is, if the time it takes for the flue gas discharged from the combustion furnace 60 to reach the flue gas concentration measuring instrument 20 is τ ₁ and the time required for the flue gas concentration measuring instrument 20 to make a measurement is τ ₂ , then τ may be equal to τ ₁ or may be equal to τ ₁ + τ _2. Furthermore, if there is any other time that affects the time delay τ, that time may also be taken into consideration. Note that τ ₁ can be calculated from the distance of the flue duct 70 from the combustion furnace 60 until the flue gas reaches the flue gas concentration measuring instrument 20 and the flow velocity of the flue gas in the flue duct 70.

The peripheral sensors 50 are various sensors that measure various physical quantities around or in the vicinity of the combustion furnace 60 and various physical quantities of equipment related to the combustion furnace 60 (e.g., the flue 70, etc.), and output the measurement values to the gas concentration estimation device 10. Hereinafter, the total number of peripheral sensors 50 is defined as n, and when the peripheral sensors 50 are not distinguished from one another, they are referred to as "peripheral sensors 50," and when they are distinguished from one another, they are referred to as "peripheral sensors 50 ₁ ,""peripheral sensors 50 ₂ ," ..., "peripheral sensors 50 _n ." Furthermore, for i = 1, 2, ..., n, the measurement value measured by the peripheral sensor 50 _i at time t is defined as x _i (t). Here, examples of physical quantities measured by the peripheral sensors 50 include the temperature inside the combustion furnace 60, the pressure inside the combustion furnace 60, the temperature inside the flue 70, the flow velocity of the exhaust gas inside the flue 70, the outside air temperature, and wind speed. However, these physical quantities are merely examples, and it goes without saying that the physical quantities measured by the surrounding sensor 50 are not limited to these.

At each time t, the gas concentration estimation device 10 receives the gas concentration y(t-τ) from the exhaust gas concentration measuring instrument 20 and also receives the physical quantities x ₁ (t), x ₂ (t), . . . , x _n (t) from the surrounding sensors 50, and calculates an estimated value of the gas concentration y(t) at that time t.

and output it to the control device 30. In the following text of the specification, the symbol "^" indicating that this is an estimated value of the gas concentration y(t) is written immediately after y, and the estimated value of the gas concentration y(t) at time t is represented as "y^(t)."

That is, the gas concentration estimation device 10 estimates the gas concentration at each time t (i.e., the gas concentration without time delay) y^(t) from the time-delayed gas concentration y(t-τ) and the physical quantities x ₁ (t), x ₂ (t), ..., x _n (t) at that time t. In other words, the gas concentration estimation device 10 functions as a software sensor that estimates the gas concentration y^(t) at each time t.

Here, the gas concentration estimation device 10 executes a model estimation process to estimate a model for estimating gas concentrations without time delay (hereinafter, this model will be represented by f and also referred to as "gas concentration estimation model f"), and a gas concentration estimation process to estimate gas concentrations without time delay using this gas concentration estimation model f. The model estimation process is a process executed offline (i.e., when the combustion furnace 60 is not operating), while the gas concentration estimation process is a process executed online (i.e., when the combustion furnace 60 is operating).

The control device 30 calculates the optimal opening/closing angle (or opening/closing amount) of the valve 40 based on the gas concentration y^(t) received from the gas concentration estimation device 10, and outputs a control command including the opening/closing angle to the valve 40. The valve 40 is a fuel valve that adjusts the amount of fuel flowing into the combustion furnace 60, and an air valve that adjusts the amount of air (oxygen) flowing into the combustion furnace 60.

Here, the heat loss and thermal efficiency within the combustion furnace 60 fluctuate depending on the amount of fuel and air flowing into the combustion furnace 60, which in turn fluctuates the gas concentration within the combustion furnace 60. Therefore, the control device 30 uses known control technology to calculate the opening/closing angle (or opening/closing amount) of the valve 40 so that the heat loss and thermal efficiency within the combustion furnace 60 are optimized, and outputs this opening/closing angle (or opening/closing amount) to the valve 40. There are various known control technologies such as this, including the technology described in Japanese Patent No. 6135831, for example.

In general, the relationship between excess air ratio and heat loss, and the relationship between excess air ratio and thermal efficiency, are shown in Figure 2. In Figure 2, the vertical axis represents heat loss or thermal efficiency, and the horizontal axis represents excess air ratio. The excess air ratio is the ratio of the amount of air actually flowing into the combustion furnace 60 to the theoretical amount of air required for combustion per unit of fuel. In Figure 2, line 1001 represents heat loss due to excess air, and curve 1002 represents heat loss due to incomplete combustion. As shown by line 1001, the greater the excess air ratio is above 1, the more excess air is heated, resulting in greater heat loss. On the other hand, as shown by curve 1002, a small excess air ratio causes incomplete combustion, resulting in greater heat loss due to CO generation, and when it exceeds a certain threshold, smoke is generated. Also in Figure 2, curve 2001 represents the thermal efficiency of the combustion furnace 60. As shown by curve 2001, thermal efficiency is greatest in region D, which includes the excess air ratio, where heat loss due to excess air and heat loss due to incomplete combustion are approximately the same, and decreases as the excess air ratio moves away from region D. Therefore, the control device 30 calculates the opening and closing angles (or opening and closing amounts) of the valves 40 (fuel valve and air valve) so that the heat loss and thermal efficiency are within region D, and outputs control commands including these opening and closing angles to the valves 40.

Note that the overall configuration of the combustion control system 1 shown in Figure 1 is an example and is not limited to this. For example, the system may include various facilities, equipment, devices, etc. that are not shown.

<Hardware Configuration of Gas Concentration Estimation Device 10>
An example of the hardware configuration of the gas concentration estimation apparatus 10 according to this embodiment is shown in Fig. 3. As shown in Fig. 3, the gas concentration estimation apparatus 10 according to this embodiment includes an input device 101, a display device 102, an external I/F 103, a communication I/F 104, a processor 105, and a memory device 106. These pieces of hardware are connected to each other via a bus 107 so as to be able to communicate with each other.

The input device 101 is, for example, a keyboard, a mouse, a touch panel, various physical buttons, etc. The display device 102 is, for example, a display, a display panel, etc. Note that the gas concentration estimation device 10 does not necessarily have to have at least one of the input device 101 and the display device 102, for example.

The external I/F 103 is an interface with external devices such as a recording medium 103a. Examples of recording media 103a include a CD (Compact Disc), a DVD (Digital Versatile Disk), an SD memory card (Secure Digital memory card), and a USB (Universal Serial Bus) memory card.

The communication I/F 104 is an interface for connecting the gas concentration estimation device 10 to a communication network. The processor 105 is, for example, a central processing unit (CPU) or a graphics processing unit (GPU). The memory device 106 is, for example, a solid state drive (SSD), random access memory (RAM), read-only memory (ROM), flash memory, or other storage device.

Note that the hardware configuration shown in FIG. 3 is an example, and the gas concentration estimation device 10 may have other hardware configurations. For example, the gas concentration estimation device 10 may have multiple processors 105 or multiple memory devices 106, or may have various types of hardware other than the hardware shown in the figure.

<Functional configuration of gas concentration estimation device 10>
An example of the functional configuration of the gas concentration estimation apparatus 10 according to this embodiment is shown in Fig. 4. As shown in Fig. 4, the gas concentration estimation apparatus 10 according to this embodiment includes a model estimation processing unit 201 and a gas concentration estimation processing unit 202. These units are realized, for example, by a processor 105 executing one or more programs installed in the gas concentration estimation apparatus 10. The gas concentration estimation apparatus 10 according to this embodiment also includes a performance data storage unit 203. The storage unit 203 is realized, for example, by the memory device 106, but may also be realized by a database server connected via a communication network.

The performance data storage unit 203 stores the gas concentration y(t) measured by the exhaust gas concentration measuring instrument 20 and the physical quantities _x1 (t), _x2 (t), ..., _xn (t) measured by each peripheral sensor 50 as performance data z(t):=(y(t), _x1 (t), _x2 (t), ..., _xn (t)). That is, the performance data storage unit 203 stores a performance data set {z(t)}, which is a collection of performance data. At each time t, the gas concentration estimation device 10 receives the gas concentration y(t-τ) with a time delay τ and the physical quantities x ₁ (t), x ₂ (t), ..., x _n (t), and this gas concentration y(t-τ) is associated with the physical quantities x ₁ (t-τ), x ₂ (t-τ), ..., x _n (t-τ) received at past times t-τ to obtain the actual data z(t-τ) = (y(t-τ), x ₁ (t-τ), x ₂ (t-τ), ..., x _n (t-τ)) at time t-τ.

The model estimation processing unit 201 estimates a gas concentration estimation model f from the actual data set {z(t)}. Here, the model estimation processing unit 201 includes a data acquisition unit 211 and a model estimation unit 212. The data acquisition unit 211 acquires the actual data set {z(t)} from the actual data storage unit 203. The model estimation unit 212 estimates a gas concentration estimation model f using the actual data set {z(t)} acquired by the data acquisition unit 211. Note that the gas concentration estimation model f is, for example, a statistical model or a machine learning model, and has a parameter θ. Hereinafter, when it is specified that the gas concentration estimation model f has the parameter θ, it will be expressed as "f _θ ".

At each time t, the gas concentration estimation processor 202 estimates the gas concentration y^(t) at the current time t from the gas concentration y(t-τ) with a time lag τ and each physical quantity x ₁ (t), x ₂ (t), ..., x _n (t) using the gas concentration estimation model f estimated by the model estimation processor 201. That is, the gas concentration estimation processor 202 calculates the estimated value y^(t) of the gas concentration y(t) at the current time t by y^(t) = _f (y(t-τ), x ₁ (t), x ₂ (t), ..., x _n (t)) using y(t-τ), x ₁ (t), x ₂ (t), ..., x n (t) as explanatory variables and y(t) as a response variable.

<Model estimation process>
The model estimation process according to this embodiment will be described with reference to Fig. 5. In the following, it is assumed that a performance data set {z(t)|z = 1, ..., T} for a certain period in the past is stored in the performance data storage unit 203. Note that in the following model estimation process, time t is a time in the past.

The data acquisition unit 211 of the model estimation processing unit 201 acquires the actual data set {z(t) | z = 1, ..., T} from the actual data storage unit 203 (step S101).

The model estimation unit 212 of the model estimation processing unit 201 estimates a gas concentration estimation model f using the actual data set {z(t)|z=1, ..., T} acquired in step S101 (step S102). That is, the model estimation unit 212 uses the actual data set {z(t)|z=1, ..., T} as learning data to learn the parameter θ so as to improve the estimation accuracy of the gas concentration y^(t)=f _θ (y(t-τ), x ₁ (t), x ₂ (t), ..., x _n (t)) at time t.

For example, if a multiple regression model is used as the gas concentration estimation model f, this gas concentration estimation model f is expressed as follows:

Here, a, b _i , and c are parameters, that is, θ=(a, b ₁ , . . . , b _n , c).

At this time, the model estimation unit 212 updates (learns) the parameter θ using, for example, a known optimization method, so as to minimize the error E between y(t) and y^(t). Note that various errors can be used as the error E, but if the mean square error is used, for example, the error E can be expressed as follows:

Although the above description has been given as an example of a case where a multiple regression model is used as the gas concentration estimation model f, various statistical models and machine learning models other than the multiple regression model can be used as the gas concentration estimation model f. For example, in addition to linear models such as the multiple regression model, nonlinear models such as support vector regression (SVR) and neural networks can also be used as the gas concentration estimation model f.

<Gas concentration estimation process>
The gas concentration estimation process according to this embodiment will be described with reference to FIG. 6. This gas concentration estimation process is repeatedly executed at each time t, but the following describes the gas concentration estimation process for a certain time t. Note that in the following gas concentration estimation process, time t is the current time. It is also assumed that the parameter θ of the gas concentration estimation model f = _fθ has already been learned in the above model estimation process.

The gas concentration estimation processing unit 202 receives the gas concentration y(t-τ) with a time lag τ and the physical quantities x ₁ (t), x ₂ (t), . . . , x _n (t) as measurement data (step S201).

The gas concentration estimation processing unit 202 uses the gas concentration estimation model f to estimate the gas concentration y^(t) at the current time t from the measurement data y(t-τ), x ₁ (t), x ₂ (t), ..., x _n (t) received in step S201 (step S202). That is, the gas concentration estimation processing unit 202 estimates the gas concentration y^(t) at the current time t by y^(t) = f(y(t-τ), x ₁ (t), x ₂ (t), ..., x _n (t)). The gas concentration y^(t) is output to the control device 30, which controls the opening/closing angle (or opening/closing amount) of the valve 40, etc. This controls the gas concentration in the combustion furnace 60.

<Summary>
As described above, in the combustion control system 1 according to this embodiment, the gas concentration estimation device 10 can estimate the gas concentration y^(t) without a time delay from the gas concentration y(t-τ) with a time delay measured by the exhaust gas concentration measuring instrument 20, which is an infrared gas analyzer. Therefore, the control device 30 can perform high-performance control of the combustion furnace 60 (that is, control that achieves stable high combustion efficiency and high energy savings) using this gas concentration y^(t) without a time delay.

Note that the gas concentration estimation device 10 according to this embodiment uses a gas concentration estimation model f realized by a statistical model or machine learning model when estimating the gas concentration at the current time t. However, for example, a pre-designed physical model may be used as the gas concentration estimation model f, and the gas concentration at the current time t may be estimated using this physical model.

The present invention is not limited to the specifically disclosed embodiments above, and various modifications, alterations, and combinations with known technologies are possible without departing from the scope of the claims.

REFERENCE SIGNS LIST 1 combustion control system 10 gas concentration estimation device 20 exhaust gas concentration measuring instrument 30 control device 40 valve 50 surrounding sensor 60 combustion furnace 70 flue 101 input device 102 display device 103 external I/F
103a Recording medium 104 Communication I/F
105 Processor 106 Memory device 107 Bus 201 Model estimation processing unit 202 Gas concentration estimation processing unit 203 Performance data storage unit 211 Data acquisition unit 212 Model estimation unit

Claims (6)

A gas concentration estimation device that estimates a gas concentration representing a CO concentration or an O2 concentration _in a combustion furnace in order to control the combustion furnace,
A receiving means for receiving a gas concentration measurement value with a time delay from the measurement time , which is measured by an infrared gas analyzer to measure the concentration of CO or _O2 emitted from the combustion furnace to the flue, and a plurality of physical quantity measurement values, which are obtained by measuring a plurality of predetermined physical quantities related to the combustion furnace by sensors;
an estimation means for calculating an estimate of the measured gas concentration value without a time delay as a gas concentration for controlling the combustion furnace from the measured gas concentration value with a time delay and the plurality of measured physical quantities using a model for estimating the gas concentration without a time delay;
and
a gas concentration estimation device, wherein the plurality of physical quantity measurement values include at least a temperature inside the combustion furnace, a pressure inside the combustion furnace, a flow velocity of gas discharged into the flue, an outside air temperature outside the combustion furnace, a wind speed outside the combustion furnace, physical quantities measured in the vicinity of or around the combustion furnace, and physical quantities measured in equipment related to the combustion furnace . the model is a statistical model or a machine learning model having parameters to be learned,
a learning means for learning parameters of the model using the gas concentration measurement values and the plurality of physical quantity measurement values over a predetermined period in the past as learning data;
The estimation means
2. The gas concentration estimation device according to claim 1, wherein an estimate of the gas concentration measurement value without a time delay is calculated as a gas concentration for controlling the combustion furnace from the gas concentration measurement value with a time delay and the plurality of physical quantity measurement values using the model having learned parameters. Let τ be the time delay, y(t) be the gas concentration measurement value at time t, n be the number of the plurality of physical quantities, and x ₁ (t), ..., x _n (t) be the physical quantity measurement values at time t.
The receiving means
At each time t, the time-delayed gas concentration measurement value y(t-τ) and the plurality of physical quantity measurement values x ₁ (t), ..., x _n (t) are received;
The estimation means
3. The gas concentration estimation device according to claim 1, wherein at each time t, an estimate of the gas concentration measurement value y(t) without the time delay is calculated from the gas concentration measurement value y(t-τ) with the time delay and the plurality of physical quantity measurement values x ₁ (t), ..., x _n (t) using the model. 4. The gas concentration estimation device according to claim 3, wherein the time delay τ is expressed as τ = τ ₁ or τ = τ ₁ + τ ₂ , where τ ₁ is the time required for the gas discharged from the combustion furnace into the flue to reach the infrared gas analyzer, and τ 2 _is the time required for the infrared gas analyzer to measure the concentration of the gas. A gas concentration estimation device that estimates a gas concentration representing a CO concentration or an O2 concentration _in a combustion furnace in order to control the combustion furnace,
a receiving step of receiving a gas concentration measurement value with a time delay from the measurement time , which is measured by an infrared gas analyzer to measure the concentration of CO or _O2 emitted from the combustion furnace into the flue, and a plurality of physical quantity measurement values, which are measured by sensors to measure a plurality of predetermined physical quantities related to the combustion furnace;
an estimation procedure for calculating an estimate of the measured gas concentration value without a time delay as a gas concentration for controlling the combustion furnace from the measured gas concentration value with a time delay and the plurality of measured physical quantities using a model for estimating the gas concentration without a time delay;
Run
a gas concentration estimation method, wherein the plurality of physical quantity measurement values include at least a temperature inside the combustion furnace, a pressure inside the combustion furnace, a flow velocity of gas discharged into the flue, an outside air temperature outside the combustion furnace, a wind speed outside the combustion furnace, physical quantities measured in the vicinity of or around the combustion furnace, and physical quantities measured in equipment related to the combustion furnace . A gas concentration estimation device that estimates a gas concentration representing a CO concentration or an O2 concentration _in a combustion furnace in order to control the combustion furnace,
a receiving step of receiving a gas concentration measurement value with a time delay from the measurement time , which is measured by an infrared gas analyzer to measure the concentration of CO or _O2 emitted from the combustion furnace into the flue, and a plurality of physical quantity measurement values, which are measured by sensors to measure a plurality of predetermined physical quantities related to the combustion furnace;
an estimation procedure for calculating an estimate of the measured gas concentration value without a time delay as a gas concentration for controlling the combustion furnace from the measured gas concentration value with a time delay and the plurality of measured physical quantities using a model for estimating the gas concentration without a time delay;
Execute
The program, wherein the plurality of physical quantity measurement values include at least the temperature inside the combustion furnace, the pressure inside the combustion furnace, the flow velocity of gas discharged into the flue, the outside air temperature outside the combustion furnace, the wind speed outside the combustion furnace, physical quantities measured in the vicinity of or around the combustion furnace, and physical quantities measured in equipment related to the combustion furnace .