JP7438014B2 - Control device - Google Patents

Description

The present disclosure relates to control of a state quantity (controlled quantity) of a controlled object in a plant, and particularly relates to determination of a manipulated variable of an operating end that can control the controlled quantity.

Boilers used in plants such as power plants measure state quantities such as main steam temperature, main steam pressure, reheat steam temperature, and NOx concentration at the outlet of the denitrification equipment, and make sure that the measured values of the state quantities become the set values. Feedback control is performed. Further, in such feedback control, the response to a sudden change in the state quantity due to the occurrence of a disturbance is delayed, and the control quantity may deviate from the permissible variation range. For this reason, when a disturbance occurs, a preliminary amount is calculated to cancel the effect before it appears on the control amount, and the feed forward is applied to the operation end by adding the operation amount to the operation end calculated by feedback control. In some cases, control is also performed together.

Disturbances to the above-mentioned main steam temperature and pressure, reheated steam temperature, NOx concentration, etc. are caused by various heat exchangers such as the furnace wall evaporation tube that forms the furnace wall and the superheater and reheater installed in the flue. This is caused by events such as the activation of a soot blower to remove ash adhering to the surface of the vessel, the ignition of a burner, and the extinguishing of a burner. Specifically, by operating the soot blower for the furnace wall evaporation tube, temperatures such as the main steam temperature and reheat steam temperature decrease, and by operating the soot blower for the heat exchanger installed in the flue, These temperatures increase. For this reason, for example, control is performed to proactively reduce or increase the spray amount in a temperature reducer for adjusting the main steam temperature within a certain range (see Patent Document 1).

Patent No. 6087226

As described above, even when the operation amount of the operating end is determined by feedback control and feedforward control, the optimal advance amount changes depending on the operating state and its change over time. Therefore, it is necessary to calculate the advance amount according to the driving state and the like.

In view of the above circumstances, it is an object of at least one embodiment of the present invention to provide a control device that is capable of more appropriately determining an advance amount for feedforward control of a control amount of a plant.

A control device according to at least one embodiment of the present invention includes:
A control device that determines the operation amount of a target operating end that can control the control amount of a plant at each prescribed timing,
Outputting a predicted variation value of the controlled variable according to the input of the process parameter using a learning model that has learned the relationship between the process parameter for specifying the operating state of the plant and the variation value of the controlled variable. a fluctuation prediction unit configured as such,
a preceding amount calculation unit configured to calculate, based on the predicted fluctuation value, a preceding amount for correcting a first manipulated variable calculated according to a deviation between a measured value of the controlled variable and a control setting value; , comprising:
The advance amount calculation unit is
a first calculation unit configured to calculate, when an event that changes the control amount, a specified preceding amount according to a duration of the event;
a second calculation unit configured to calculate a leading indicator for correcting the prescribed leading amount based on the predicted fluctuation value and the deviation;
and a third calculation unit configured to calculate the lead amount based on the prescribed lead amount and the lead indicator.

According to at least one embodiment of the present invention, a control device is provided that is capable of more appropriately determining an advance amount for feedforward control of a control amount of a plant.

1 is a diagram schematically showing the configuration of a boiler included in a plant 7 according to an embodiment of the present invention. FIG. 1 is a block diagram showing the configuration of a control device according to an embodiment of the present invention. 3 is a diagram showing the configuration of a first calculation section and a third calculation section according to an embodiment of the present invention, and is a detail of the portion indicated by A in FIG. 2. FIG. FIG. 3 is a diagram illustrating a relationship between a prescribed advance amount calculated by a first calculation unit and an event duration time according to an embodiment of the present invention. FIG. 3 is a diagram illustrating a relationship between a prescribed leading indicator value and a predicted fluctuation value calculated by a prescribed leading indicator calculation unit according to an embodiment of the present invention. FIG. 2 is a diagram showing the configuration of an evaluation section according to an embodiment of the present invention.

Hereinafter, some embodiments of the present invention will be described with reference to the accompanying drawings. However, the dimensions, materials, shapes, relative arrangements, etc. of the components described as the embodiments or shown in the drawings are not intended to limit the scope of the present invention thereto, and are merely illustrative examples. do not have.
For example, expressions expressing relative or absolute positioning such as "in a certain direction,""along a certain direction,""parallel,""orthogonal,""centered,""concentric," or "coaxial" are strictly In addition to representing such an arrangement, it also represents a state in which they are relatively displaced with a tolerance or an angle or distance that allows the same function to be obtained.
For example, expressions such as "same,""equal," and "homogeneous" that indicate that things are in an equal state do not only mean that things are exactly equal, but also have tolerances or differences in the degree to which the same function can be obtained. It also represents the existing state.
For example, expressions expressing shapes such as squares and cylinders do not only refer to shapes such as squares and cylinders in a strict geometric sense, but also include uneven parts and chamfers to the extent that the same effect can be obtained. Shapes including parts, etc. shall also be expressed.
On the other hand, the expressions "comprising,""comprising,""comprising,""containing," or "having" one component are not exclusive expressions that exclude the presence of other components.

FIG. 1 is a diagram schematically showing the configuration of a boiler 71 included in a plant 7 according to an embodiment of the present invention.
The control device 1 periodically controls a manipulated variable MV (command value) of an operating end (hereinafter referred to as a target operating end 9) that can control the state quantity (hereinafter referred to as a controlled amount) of a plant 7 that is a control target of the control device 1. This is a device that can determine each specified timing such as. In general, the plant 7 includes a plurality of operating terminals that can change a movable part such as a valve body (valve opening degree) according to a manipulated variable MV given from the outside, and various control terminals such as temperature, pressure, flow rate, etc. A plurality of sensors are installed to measure state quantities. The control device 1 determines the operation amount MV of the target operation end 9, which is one or more of the plurality of operation ends installed in the plant 7, so that the measured value PV of the control amount obtained using the above-mentioned sensor is the control setting value. It is determined by feedback control and feedforward control so that the SV (target value) is achieved. Further, the control device 1 can control the target operation end 9 by outputting (transmitting) the determined operation amount MV to the outside.

The plant 7 shown in FIG. 1 is a power generation plant that includes a boiler 71 such as a boiler turbine generator (BTG), for example. The power generation plant rotationally drives the generator G connected to the rotating shaft of the steam turbine T by supplying superheated steam (main steam) generated in the boiler 71 to, for example, the steam turbine T and rotationally driving the rotating shaft of the steam turbine T. power generation. In the plant 7 including such a boiler 71, the control device 1 may use at least one of the state variables such as the main steam temperature, the main steam pressure, the reheat steam temperature, and the NOx concentration at the outlet of the denitrification device as the controlled variable. . The target operation end 9 for controlling these control amounts is a spray valve 7v, a fuel supply device 72v, a gas recirculation fan inlet damper 87 installed at the inlet of a gas recirculation fan 86, or a burner, respectively, as described later. These include an angle adjuster (not shown), an ammonia flow control valve 81v, and the like.

In the embodiment shown in FIG. 1, the boiler 71 exchanges heat generated by burning pulverized coal fuel, which is obtained by pulverizing coal, inside the furnace 71f with feed water and steam, and supplies the heat to a steam turbine T, etc. This is a pulverized coal-fired boiler configured to generate superheated steam (main steam). However, the boiler 71 may be a waste heat recovery boiler such as a soda recovery boiler that uses different fuels or combustion methods, a fluidized bed boiler, a fluidized bed boiler, or the like.

First, the boiler 71 will be explained in detail using the pulverized coal boiler shown in FIG. 1 as an example. As shown in FIG. 1, the boiler 71 (boiler main body) includes a furnace 71f, a combustion device 72 installed in the furnace 71f, and a flue 73 connected to the boiler 71.

The furnace 71f has a combustion chamber therein, has the hollow shape of, for example, a rectangular cylinder, and is installed along the vertical direction (see FIG. 1). The inner wall surface of the wall of the furnace 71f (furnace wall) forming the combustion chamber is composed of a furnace wall evaporation tube 71p (heat transfer tube) for generating steam from water and a fin connecting the furnace wall evaporation tube 71p. Ru. Specifically, for example, a plurality of furnace wall evaporation tubes 71p arranged along the vertical direction are arranged horizontally, and the adjacent furnace wall evaporation tubes 71p and the furnace wall of the plurality of furnace wall evaporation tubes 71p are arranged horizontally. The fins are arranged so as to close the space between the evaporation pipe 71p and the evaporation pipe 71p. The furnace 71f is provided with a sloped surface at the bottom of the furnace, and a furnace wall evaporation tube 71p (bottom evaporation tube) is also provided on the sloped surface, which becomes the bottom surface. Further, inside the combustion chamber, an ash removal device such as a soot blower (not shown) configured to remove deposits such as ash and unburned matter adhering to the surface of the furnace wall evaporation tube 71p is provided.

The combustion device 72 is a device for supplying fuel to the inside of the furnace 71f for combustion, and supplies fuel to one or more combustion burners 74 installed on the furnace wall, the wind box 77e, and the boiler 71. A fuel supply device 72v for adjusting the amount is provided. Specifically, the furnace 71f is configured to combust fuel by supplying fuel through the combustion burner 74 and supplying combustion air (secondary air) through the wind box 77e. Ru. The amount of fuel supplied to the combustion burner 74 is adjusted by controlling the fuel supply device 72v (operating end) so that the measured value PV (control amount) of the main steam pressure becomes the control set value SV. . In the embodiment shown in FIG. 1, the steam generated in the furnace wall evaporation pipe 71p is superheated in the superheater 79a as it flows through the steam pipe 7p, and then sent to the steam turbine T. The pressure is detected by a pressure sensor 7s installed at the boiler outlet.

As shown in FIG. 1, this combustion device 72 may further include an AA port 77p installed in the furnace wall for supplying additional air (AA). In this case, combustion air is generated by conveying air (primary air) to be described later for supplying fuel to the combustion burner 74 and combustion air (secondary air) supplied to the furnace 71f via the wind box 77e. By setting the supply amount to be less than the theoretical air amount relative to the supply amount of pulverized coal fuel, the interior is maintained in a reducing atmosphere, and after reducing and burning the pulverized coal fuel, new combustion air is added. (AA) is supplied to complete the oxidation combustion of the pulverized coal fuel, making it possible to reduce the amount of NOx generated on the combustion burner 74 side (two-stage combustion method).

In this embodiment, the combustion device 72 includes a plurality of combustion burners 74. One or more combustion burners 74 (in FIG. 1, a plurality of combustion burners) are installed on the furnace wall at equal intervals, for example, in the circumferential direction of the furnace 71f or in positions facing the furnace 71f. One or more combustion burners 74 are installed along the vertical direction in one or more stages (five stages 74a to 74e in FIG. 1). These plurality of combustion burners 74 are each connected to one of a plurality of pulverizers 76 (76a to 76e in FIG. 1; pulverizer/mill) via a pulverized coal supply pipe 75, and each of the pulverizers 76 The pulverized coal produced is supplied to the combustion burner 74 by conveying air (primary air) via the pulverized coal supply pipe 75 in which the above-mentioned fuel supply device 72v is installed. The above-mentioned crusher 76 is configured to generate pulverized coal by pulverizing coal supplied by a raw coal conveyance system (not shown) into a predetermined size of fine powder.

Moreover, the above-mentioned wind box 77e is provided at the installation position of each combustion burner 74 in the furnace 71f, and is connected to one end of an air duct 77a that guides air (combustion air). The other end of the air duct 77a is connected to a blower 78, so that air pushed by the blower 78 is supplied to the wind box 77e through the air duct 77a. In addition, a branch duct 77c branches from a branch part 77b provided in the air duct 77a and is connected to the above-mentioned AA port 77p, so that AA is controlled according to the opening degree of the AA damper 77d installed in the branch duct 77c. It is configured such that it can be supplied to the AA port 77p (inside the furnace 71f). In other words, the air supplied from the blower 78 to the air duct 77a is heated by the air heater 82 (described later), and then the secondary air is guided to the wind box 77e at the branch part 77b, and the air is passed through the branch duct 77c to the AA port. It is distributed to AA which leads to 77p.

The flue 73 is a duct that guides combustion gas generated by combustion of fuel in the furnace 71f, and is connected vertically above the furnace 71f, as shown in FIG. This flue 73 includes a plurality of heat exchangers 79 (in FIG. 1, a superheater 79a (SH), a reheater 79b (RH), and a energy saver) for generating steam and heating (superheating) the steam. Ash removal devices (not shown) are installed for each heat exchanger 79 in order to remove deposits (ash, etc.) adhering to the heat transfer surfaces of the heat exchangers 79 and 79c (ECO). Ru. Then, the combustion gas flowing into the flue 73 from the furnace 71f heats the feed water and steam flowing through the furnace wall and the heat exchanger 79 to generate main steam and the like.

More specifically, in the case of a once-through boiler, for example, a gas-liquid mixed fluid in which water (saturated water) and steam (saturated steam) are mixed is produced by boiling water inside the furnace wall evaporation tube 71p. The gas-liquid mixed fluid flows into a steam (brackish water) separator (not shown) and is separated into steam (gas phase) and water (liquid phase). After that, the steam (main steam) separated from water in the moisture (brackish water) separator is sent to the steam turbine T etc. through the steam pipe 7p (superheated steam pipe) that forms the steam flow path. The main steam temperature (control amount) is kept constant at the control set value SV by heating by the superheater installed in the steam pipe 7p and cooling as necessary by the temperature reducer 7c (superheat reducer). Adjustments will be made to

In the embodiment shown in FIG. 1, a coolant supply pipe 7L that supplies a coolant 7w is connected to the temperature reducer 7c, and the spray valve 7v (operating end) supplies the coolant to the temperature reducer 7c. The flow rate of the cooling medium 7w is adjustable. Then, the valve opening degree of the spray valve 7v (operating end) is adjusted so that the measured value PV (control amount) of the thermometer (boiler outlet temperature measurement sensor 7t) installed in the steam pipe 7p becomes the control set value SV. It is supposed to be done.
On the other hand, regarding the steam temperature heated by the reheater 79b (reheat steam temperature), the gas recirculation fan inlet damper 87 (operating end) and a burner angle adjuster (not shown) that can adjust the burner nozzle angle (operating end).

Further, by connecting an exhaust gas treatment device to the downstream side of the flue 73, the combustion gas (exhaust gas) after passing through the heat exchanger 79 is rendered harmless by the exhaust gas treatment device and then released into the atmosphere. released. More specifically, the exhaust gas treatment device includes an exhaust gas passage 8 connected to the flue 73 and a plurality of devices installed in the exhaust gas passage 8. In the embodiment shown in FIG. 1, the exhaust gas passage 8 includes, in order from the upstream side to the downstream side of the exhaust gas, air supplied to the air duct 77a from the denitrification device 81 and the blower 78, and exhaust gas flowing through the exhaust gas passage 8. An air heater 82, a soot and dust treatment device 83, an induced blower 84, etc., which exchange heat with the air heater 82, and the like are provided, and a chimney 85 is provided at the downstream end.

The denitration device 81 described above purifies NOx contained in the exhaust gas passing through the catalyst by supplying 8w of ammonia or the like to the catalyst installed inside. Specifically, a pipe (ammonia supply pipe 81p) for supplying ammonia 8w is connected to the denitrification device 81, and by adjusting the valve opening degree of the ammonia flow control valve 81v (operating end), The amount of ammonia 8w supplied to the denitrification device 81 can be controlled. A sensor (NOx sensor 81s) capable of detecting NOx concentration is installed at the outlet of the denitrification device 81, and an ammonia flow control valve 81v is set so that the measured value PV by the NOx sensor 81s becomes the control set value SV. (operating end) is controlled.

As described above, the plant 7 includes a plurality of devices, each of which includes a plurality of devices, such as the boiler 71, an exhaust gas treatment device, a steam turbine T, an operating end, and a sensor. Then, the control amount is controlled by the control device 1 or the like so that the operation of the plant 7 is optimal while satisfying the output request based on, for example, electric power demand of the plant 7 (boiler 71). Specifically, the control device 1 and the like performs feedforward control in addition to feedback control.

Here, the feedforward control is performed in order to suppress the influence of the occurrence of an event that causes a disturbance in the feedback control of the control amount. For example, a manipulated variable (hereinafter referred to as a first manipulated variable Fb) determined by feedback control is determined at a specified timing for the spray valve 7v so that the measured value PV of the main steam temperature becomes the control set value SV. At this time, when the ash removal device of the furnace wall evaporation pipe 71p operates (occurrence of an event), the deposits are removed, and the heat generated by combustion in the furnace 71f and the fluid (water) inside the evaporation pipe 71p are removed. The amount of heat exchanged between the This becomes a disturbance to the feedback control, and the temperature of the combustion gas flowing toward the flue 73 decreases accordingly, causing a rapid decrease in main steam temperature and a rapid increase in steam pressure. In addition, when the ash removal device of the superheater 79a operates (occurrence of an event), the amount of heat exchange increases rapidly, which causes a disturbance to the feedback control, causing excessive heating of the main steam, and the temperature of the main steam. leading to a rapid increase in

Similarly, the first manipulated variable Fb for at least one of the gas recirculation fan inlet damper 87 or the burner angle adjuster (not shown) is adjusted at a specified timing so that the measured value PV of the reheat steam temperature becomes the control set value SV. It is determined. At this time, the operation of the ash removal device of the furnace wall evaporation pipe 71p (occurrence of an event) causes the above-mentioned disturbance, and the temperature of the reheated steam rapidly decreases. Further, the operation of the ash removal device of the reheater 79b causes the above-mentioned disturbance, and the temperature of the reheated steam rises rapidly.

Regarding the main steam pressure, the first manipulated variable Fb for the fuel supply device 72v is determined at a specified timing so that the measured value PV of the main steam pressure becomes the control set value SV. (occurrence of an event) causes the above-mentioned disturbance, and the main steam pressure rises rapidly.

Regarding the NOx concentration at the outlet of the denitrification device 81, the first operation amount Fb for the ammonia flow control valve 81v is determined at a specified timing so that the measured value PV of the NOx concentration becomes the control set value SV. The ignition or extinguishing of the combustion burner 74 (occurrence of an event) causes the above-mentioned disturbance, and the NOx concentration of the steam rapidly increases.

Then, in response to the disturbance in the feedback control of the control amount as described above, a manipulated variable (hereinafter referred to as advance amount Ff) to be given in advance according to the occurrence of the disturbance is calculated by feedforward control, and its optimal value is as follows. It changes depending on the operating state of the plant 7 (boiler 71) and its changes over time. Therefore, in order to calculate an appropriate advance amount Ff according to the operating state of the plant 7 (boiler 71), the control device 1 calculates (computes) the advance amount Ff as described below.

Hereinafter, the above control device 1 will be explained in detail using FIGS. 2 to 6.
FIG. 2 is a block diagram showing the configuration of the control device 1 according to an embodiment of the present invention. FIG. 3 is a diagram showing the configuration of the first calculation section 31 and the third calculation section 5 according to an embodiment of the present invention, and shows details of the portion indicated by A in FIG. 2. FIG. 4 is a diagram illustrating the relationship between the specified advance amount Qa calculated by the first calculation unit 31 and the event duration t according to an embodiment of the present invention. FIG. 5 is a diagram illustrating the relationship between the prescribed leading indicator value Q1 calculated by the prescribed leading indicator calculation unit 42 and the predicted fluctuation value V according to an embodiment of the present invention. Moreover, FIG. 6 is a diagram showing the configuration of the evaluation section 43 according to an embodiment of the present invention.

As shown in FIG. 2, the control device 1 includes a fluctuation prediction section 2 and a preceding amount calculation section 3. Each of these functional units (circuits) will be explained.
Note that the control device 1 may be configured by a computer. It includes a processor 11 such as a CPU, and a storage section 12 such as a memory such as ROM or RAM. The processor 11 operates (data calculations, etc.) according to the instructions of the program (control program 10) loaded into the memory (main storage unit), thereby realizing each of the above-mentioned functional units. For example, it may be implemented using an MCU (Micro Controller Unit) or the like, or may be implemented on a single integrated circuit (chip) such as an SoC (System on Chip). Further, the above program is software for causing a computer to implement each functional unit described later, and may be stored in a storage medium readable by the computer.

The fluctuation prediction unit 2 uses a learning model M that has learned the relationship between the process parameter P for specifying the operating state of the plant 7 and the fluctuation value of the controlled variable, and calculates the controlled variable according to the input of the process parameter P. This is a functional unit configured to output a predicted fluctuation value V. The process parameter P includes at least one of the measured value of the state quantity and the manipulated variable of the operating end. The measured value of the state quantity can be measured by a sensor that is shown or not shown, and the operation amount of the operating end can be obtained from the operating end or a control section of the operating end.

For example, the process parameters P are a) main steam pressure, b) exhaust gas temperature at the outlet of the economizer 79c, c) valve opening degree of the spray valve 7v, d) fuel coal flow rate (ton/hour), e) feed water flow rate. (tons/hour), f) Main steam temperature, g) Previous leading indicator Qb (described later), h) Generator output (load index), i) Reheat steam temperature, j) Superheater 79a (primary superheater K) Outlet steam temperature of superheater 79a (primary superheater, etc.), l) Opening degree of gas recirculation fan inlet damper 87, m) Burner nozzle angle of combustion burner 74 (burner angle adjuster), setting angle), n) fuel gas flow rate, o) fuel oil flow rate, p) main steam flow rate, q) total air flow rate (total flow rate flowing into the above-mentioned branch part 77b), r) outlet temperature of the reheater 79b at least one parameter value of. Note that c), g), l), and m) above are the manipulated variables of the operating end, and the others are state variables.

In addition, the above learning model M learns the relationship between a process parameter P including the parameter values of one or more of the parameters described above and the fluctuation value of the controlled variable measured under this process parameter P (machine learning). Created by This learning may be performed using a well-known machine learning method on learning data (teacher data). The learning data includes, for example, each parameter value of the process parameter P obtained when the plant 7 (boiler 71) was operated (actual operation, test run) in the past, and the measurement result of the fluctuation value of the controlled variable when this was obtained. It may be composed of a plurality of pieces of data that are associated with each other. Alternatively, the learning data may be composed of a plurality of data in which each parameter value of the process parameter P obtained from literature or numerical analysis (material simulation) is associated with a fluctuation value of the controlled variable. These data may be combined. The learning model M created in this way outputs a variation value of the controlled variable obtained from the above relationship as a predicted variation value V for the input process parameter P.

The advance amount calculation unit 3 calculates a advance amount for correcting the first manipulated variable Fb, which is calculated according to the deviation D between the measured value PV of the controlled variable and the control set value SV, based on the predicted fluctuation value V described above. This is a functional unit configured to calculate Ff. The first manipulated variable Fb is a feedback manipulated variable calculated by feedback control based on the deviation D described above. The value corrected by the advance amount Ff is output as the final manipulated variable MV to the target operating end 9 operated by this manipulated variable MV (see FIG. 2).

To be more specific, as shown in FIG. 2, the preceding amount calculating section 3 includes a first calculating section 31 (see FIG. 3), a second calculating section 4, in order to calculate the preceding amount Ff. It has three functional units (circuits) including a third calculation unit 5 (see FIG. 3). These functional units will be explained using an example in which the event is the operation of the ash removal device, the control amount is the main steam temperature, and the target operating end 9 is the spray valve 7v.

The first calculation unit 31 calculates a specified advance amount Qa according to the duration t of the event when an event that changes the control amount occurs, such as the above-mentioned operation of the ash removal device, ignition of the combustion burner 74, and extinguishing the fire. This is a functional unit (function generator) configured as follows. More specifically, as shown in FIG. 3, the operating time of the ash removal device (the soot blower in this embodiment) of the superheater 79a is input to the first calculation unit 31 as the event duration t. Then, by using a relationship such as a function (see Fig. 4) or a table created based on control theory or experience, which can uniquely calculate the specified advance amount Qa from the input event duration t, the specified timing is calculated. A prescribed advance amount Qa is calculated according to the duration t of the event that advances each time.

In the embodiments shown in FIGS. 2 to 6, as shown in FIG. 3, the first calculation unit 31 uses a plurality of mutually different functions (three in FIG. 3) to calculate the duration t of the input event. The prescribed advance amount Qa is calculated respectively.

The second calculation unit 4 calculates the prescribed advance amount Qa based on the predicted fluctuation value V output from the fluctuation prediction unit 2 described above and the deviation D (for example, PV−SV) used in the feedback control described above. This is a functional unit configured to calculate a leading indicator Qb for correcting. That is, the second calculation unit 4 uses the predicted fluctuation value V and the deviation D as input values, and outputs the calculation result of the leading indicator Qb.

More specifically, in some embodiments, as shown in FIG. A prescribed leading indicator calculation unit 42 (function generator) that calculates a prescribed leading indicator value Q1, and an evaluation unit that calculates an adjustment value Q2 for adjusting the prescribed leading indicator value Q1 based on the deviation D. 43 (see FIG. 6), and a leading indicator calculation unit 44 that calculates the leading indicator Qb based on the prescribed leading indicator value Q1 and the adjustment value Q2.

The prescribed leading indicator calculation unit 42 may perform each calculation using a relationship such as a function that can calculate a unique value from input values, as shown in FIG. In FIG. 5, the horizontal axis is the main steam temperature fluctuation value (°C) as the predicted fluctuation value V, and the other axis is the specified leading indicator value Q1. Using a function showing such a relationship, the input A predetermined leading indicator value Q1 is calculated according to the predicted fluctuation value V.

Further, the above adjustment value Q2 is determined based on the deviation D obtained as a result of actually applying the preceding amount Ff calculated based on the predicted fluctuation value V described above to the target operating end 9, and the preceding amount Ff applied immediately before. is evaluated, and has a value smaller than the specified leading indicator value Q1 (Q1>Q2). By adjusting the specified leading indicator value Q1 using the adjustment value Q2, it has the role of adjusting the leading indicator Qb to a more appropriate value. In short, instead of directly calculating the leading indicator Qb based on the predicted fluctuation value V obtained from the learning model M, the prescribed leading indicator value Q1 calculated based on this predicted fluctuation value V is calculated immediately before that. Adjustment is performed using feedback control based on the result of applying the advance amount Ff to the target operation end 9.

Specifically, as shown in FIG. 6, the evaluation unit 43 calculates a specified adjustment value Q3 having a value corresponding to the input deviation D, which is the basis of the adjustment value Q2. The calculation unit 43a (function generator) is connected to the downstream stage of the specified adjustment value calculation unit 43a, and outputs the specified adjustment value Q3 to the downstream stage only when a specified event occurs (in FIG. 6, when the SH soot blower is operating). an output switching section 43b; an integrated value calculating section 43c connected to the downstream of the output switching section 43b and calculating an integrated value I by integrating the specified adjustment value Q3 inputted at each specified timing; The adjustment value calculation unit 43d (function generator) may be provided to calculate the adjustment value Q2. The prescribed adjustment value calculation unit 43a and the adjustment value calculation unit 43d may each perform their calculations using a relationship such as a function that can calculate a unique value from an input value.

In the embodiments shown in FIGS. 2 to 6, as shown in FIG. 6, when the event occurrence signal E indicates ON, the output switching unit 43b converts the output value of the specified adjustment value calculation unit 43a into the subsequent integrated value. It is output to the value calculation unit 43c, and when indicating OFF, 0 (%) is output to the subsequent stage. In this way, by calculating the integrated value I, instead of calculating the leading indicator Qb by modifying the function of the specified leading indicator calculation unit 42 (FIG. 2), an appropriate leading indicator can be calculated by calculating the adjustment value Q2. Calculation of the index Qb becomes possible.

Furthermore, in the embodiments shown in FIGS. 2 to 6, as shown in FIG. The index calculating section 42 is configured to calculate the prescribed leading index value Q1 by inputting the predicted fluctuation value V held in the holding section 41. This holding unit 41 is provided to hold the predicted fluctuation value V immediately before the event occurs. Since the predicted fluctuation value V also changes due to a transient change in the state quantity during the occurrence of an event, the holding unit 41 allows the advance amount Ff to be determined using the predicted fluctuation value V immediately before the occurrence of the event.

The third calculation unit 5 is a functional unit for calculating the lead amount Ff based on the specified lead amount Qa and the lead indicator Qb described above (see FIG. 3). In other words, the leading amount Ff is not directly calculated based on the predicted fluctuation value V obtained from the learning model M, but is calculated using the leading indicator Qb calculated based on the predicted fluctuation value V by the second calculation unit 4. , the predetermined predetermined amount Qa calculated by the first calculation unit 31 is corrected to a more appropriate value, thereby calculating the predetermined amount Ff. That is, the third calculation unit 5 outputs the leading amount Ff using the specified leading amount Qa and the leading indicator Qb as input values.

In the embodiments shown in FIGS. 2 to 6, as shown in FIG. The leading indicator Qb calculated by the unit 4 is input. Then, in the third calculation unit 5, one of the plurality of input predetermined advance amounts Qa is selected depending on the value of the input leading indicator Qb, and the selected predetermined advance amount Qa is It is output from the third calculation unit 5. In addition, in this embodiment, the advance amount Ff is calculated in consideration of the load value of the boiler 71, but this will be described later.

The advance amount Ff outputted from the advance amount calculation section 3 in this manner is input to the feedback control section 6, and is added to the above-mentioned first manipulated variable Fb in the addition section 63. This addition result is transmitted (output) to the target operating end 9 as the manipulated variable MV. When the target operating end 9 is operated with the manipulated variable MV, the measured value PV of the controlled variable changes, and the measured value PV of the controlled variable and the control set value SV after this change are input to the deviation calculation unit 61, The above deviation D is input to the controller 62. The controller 62 calculates the first operation amount Fb of the target operation end 9 so that the deviation D becomes 0, and outputs the calculation result to the above-mentioned addition section 63. In addition, although the controller 62 performs PI control in FIG. 2, it may perform PID control. Then, the adder 63 calculates the manipulated variable MV based on the first manipulated variable Fb and the preceding amount Ff. Feedback control is performed by repeating such control. Further, the deviation D calculated by the deviation calculation section 61 described above is input to the third calculation section 5 when the event occurrence signal E is on.

According to the above configuration, the preceding amount Ff for correcting the first manipulated variable Fb determined by feedback control of the controlled variable of the plant 7 is set to the operating data obtained from past operations of the plant 7 (actual operation, trial operation), etc. It is calculated based on the predicted fluctuation value V obtained by the learning model M that has learned the relationship between the process parameter P and the fluctuation value of the controlled variable. Specifically, the prescribed advance amount Qa, which is determined based on control theory and experience, is determined based on the duration of an event such as the operation of the ash removal device, the ignition or extinguishing of the combustion burner 74, and the predicted fluctuation value V. The leading amount Ff is calculated by correcting it using the leading indicator Qb calculated based on the deviation D between the measured value PV of the controlled amount and the control set value SV. As a result, it is possible to obtain a more appropriate advance amount Ff according to the operating state of the plant 7 and its change over time, and by optimizing the operation amount of the target operating end 9, it is possible to more appropriately calculate the control amount. .

Next, the relationship between the process parameter P and the control amount described above will be explained in detail.
In some embodiments, the process parameter P includes a measured value of one or more state variables, a preset operating variable of one or more operating terminals (C), l), m)), and a leading indicator Qb. (g))). Then, the above-mentioned fluctuation prediction unit 2 calculates a predicted fluctuation value V based on such a process parameter P.

Specifically, when the controlled variable is the main steam temperature, the process parameter P may include the above b), c), d), f), g), h), and j). Then, the advance amount calculation unit 3 calculates the advance amount Ff for the spray valve 7v, which is the target operation end 9, based on the predicted fluctuation value V of the main steam temperature. Further, this calculation result is input to the feedback control section 6 for the valve opening degree of the spray valve 7v. Thereby, feedforward control can be performed more appropriately so that a sudden rise or fall in the main steam temperature due to the above-mentioned disturbance is suppressed.

When the controlled variable is the reheat steam temperature, the process parameters P are the above b), c), d), f), g), h), i), l), m), and r). may also be included. Based on the predicted fluctuation value V of the reheat steam temperature, the advance amount calculation unit 3 determines the opening degree of the gas recirculation fan inlet damper 87, which is the target operation end 9, and the burner nozzle by the burner angle adjuster (not shown). The leading amount Ff for each angle is calculated. Further, this calculation result is input to the feedback control section 6 for each of the opening degree of the gas recirculation fan inlet damper 87 and the burner nozzle angle. Thereby, feedforward control can be performed more appropriately so that a sudden increase or decrease in the reheated steam temperature due to the above-mentioned disturbance is suppressed.

When the controlled variable is the main steam pressure, the process parameters P are the above a), b), c), d), f), g), h), j), k), l), m ), and p). Then, the advance amount calculation unit 3 calculates the advance amount Ff for the boiler master (fuel supply device 72v), which is the target operating end 9, based on the predicted fluctuation value V of the main steam pressure. Further, this calculation result is input to the valve opening degree feedback control section 6 of the fuel supply device 72v. Thereby, feedforward control can be performed more appropriately so that a sudden increase in main steam pressure due to the above-mentioned disturbance is suppressed.

Furthermore, when the control amount is the NOx concentration at the outlet of the denitrification device 81, the process parameters P are the above d), h), g), l), m), n), o), p), and q) may also be included. Then, the advance amount calculation unit 3 calculates the advance amount Ff for the ammonia flow control valve 81v, which is the target operation end 9, based on the predicted NOx concentration fluctuation value V described above. Further, this calculation result is input to the feedback control section 6 for the valve opening degree of the ammonia flow rate control valve 81v. As a result, feedforward control can be performed more appropriately so as to suppress a rapid increase in NOx concentration due to the above-mentioned disturbance.

Note that the fluctuation prediction unit 2 and the advance amount calculation unit 3 calculate the advance amount Ff at each prescribed timing, but as shown in FIG. 3, when the event duration t is 0, the prescribed advance amount Qa is Since it becomes 0, the advance amount Ff also becomes 0.

Next, an embodiment will be described in which the above-mentioned advance amount Ff is calculated in consideration of the load on the boiler 71.
In some embodiments, as shown in FIG. 2, the above-mentioned advance amount Ff may be calculated in consideration of the load on the boiler 71.

Specifically, in some embodiments, as shown in FIG. 2, the control device 1 further includes a load acquisition unit 32 that acquires the load value of the boiler 71 included in the plant 7. The above-mentioned fluctuation prediction unit 2 acquires the predicted fluctuation values V from each of the plurality of learning models M provided according to the plurality of load bands of the boiler 71 by the load acquisition unit 32 at the time of prediction. The predicted fluctuation value V may be configured to be output according to the load value of the boiler 71 (hereinafter referred to as the current load value L). Further, the above-mentioned advance amount calculation section 3 has the above-mentioned first calculation section 31, second calculation section 4, and third calculation section 5, which are respectively provided according to the plurality of load bands, and includes the above-mentioned first calculation section 31, second calculation section 4, and third calculation section 5. It may be configured to calculate the advance amount Ff according to the load value L.

More specifically, each of the plurality of learning models M is based on the above-mentioned relationship (learning data) obtained when the boiler 71 is operated at a load value representing each load zone (hereinafter referred to as representative load value). ) is created by learning. In the embodiment shown in FIG. 2, the load of the boiler 71 is divided into four load bands, and 30%, 50%, 75%, and 100% belonging to each load band are respectively set as representative load values, resulting in a total of 4 load bands. Two learning models M have been created.

Further, the above-mentioned process parameters P are input to each of the plurality of learning models M in parallel. Then, based on the output values obtained from each of the plurality of learning models M for the input value of the process parameter P and the current load value L obtained by the load obtaining section 32, the fluctuation prediction section 2 A predicted fluctuation value V to be output to the second calculation unit 4 is calculated.

Specifically, when the current load value L and the representative load value are the same, the output of the learning model M created with the representative load value having the same value as the current load value L is output as the predicted fluctuation value V. Ru. If the current load value L and the representative load value are not the same, the current load value L is A calculated value corresponding to is determined and outputted as a predicted fluctuation value V. In the embodiment shown in FIG. 2, for example, two learnings are performed in which representative load values between the current load value L (e.g., 90%) are L1 (e.g., 75%) and L2 (e.g., 100%), respectively. Assuming that the output values of model M are V1 and V2, respectively, the predicted fluctuation value V corresponding to the current load value L is located on the straight line connecting the two coordinate points (V1, L1) and (V2, L2). become.

Similarly, as shown in FIG. 2, the second calculation unit 4 and the third calculation unit 5 are configured in the same number as the number of learning models M, depending on the load bands into which the load value is divided in the same way as the learning models M. Each is provided. Specifically, the predicted fluctuation value V outputted from the fluctuation prediction section 2 is inputted to a plurality of second calculation sections 4 provided for each load zone, and the prescribed leading indicator value Q1 is input from each of them. is output. These plurality of second calculation units 4 calculate the specified leading indicator value Q1 according to the load band using mutually different functions.

Further, as shown in FIG. 2, the evaluation section 43 is provided for each of the plurality of second calculation sections 4, and the prescribed leading indicator value Q1 calculated for each load zone is calculated for each load zone. They are each adjusted by the adjusted adjustment value Q2. At this time, each evaluation section 43 calculates an adjustment value Q2 according to the load band, and the adjusted specified leading indicator value Q1 and current load value L obtained for each load band are input to the switching section 45. . Then, the switching unit 45 performs linear interpolation in the same manner as described above, and thereby outputs the leading indicator Qb.

This leading indicator Qb is input to each of the plurality of third calculation units 5 provided for each load zone. Further, the first calculation unit 31 provided for each third calculation unit 5 is configured to calculate the prescribed advance amount Qa according to the load band, and is connected to the subsequent stage of the third calculation unit 5. The switching unit 52 outputs the advance amount Ff according to the load band.

According to the above configuration, the advance amount Ff is calculated in consideration of the load of the plant 7 (boiler 71) at the time of calculation. With this, it is possible to obtain an appropriate advance amount Ff depending on the load.

The present invention is not limited to the embodiments described above, and also includes forms in which modifications are added to the embodiments described above, and forms in which these forms are appropriately combined.

<Additional notes>
(1) A control device (1) according to at least one embodiment of the present invention includes:
A control device (1) that determines a manipulated variable (MV) of a target operating end (9) capable of controlling a controlled variable of a plant (7) at each prescribed timing,
A learning model (M) that has learned the relationship between the process parameter (P) for specifying the operating state of the plant (7) and the fluctuation value of the controlled variable is used to input the process parameter (P). a fluctuation prediction unit (2) configured to output a predicted fluctuation value (V) of the controlled variable according to the control amount;
for correcting a first manipulated variable (Fb) calculated according to a deviation (D) between a measured value (PV) of the controlled variable and a control set value (SV) based on the predicted fluctuation value (V); a leading amount calculation unit (3) configured to calculate a leading amount (Ff) of;
The preceding amount calculation unit (3)
a first calculation unit (31) configured to calculate a prescribed advance amount (Qa) according to a duration (t) of the event when an event that changes the control amount occurs;
a second calculation unit (4) configured to calculate a leading indicator (Qb) for correcting the prescribed leading amount (Qa) based on the predicted fluctuation value (V) and the deviation (D);
and a third calculation unit (5) configured to calculate the preceding amount (Ff) based on the specified preceding amount (Qa) and the leading indicator (Qb).

According to the configuration (1) above, the preceding amount (Ff) (feedforward manipulated variable) that corrects the first manipulated variable (Fb) found by feedback control of the controlled variable (state variable of the controlled object) of the plant (7) is obtained by a learning model (M) that has learned the relationship between the process parameter (P) and the fluctuation value of the controlled variable from the operation data obtained from past plant (7) operations (actual operation, trial operation), etc. It is calculated based on the prediction result (V) of the fluctuation value of the controlled variable. For example, in a plant (7) with a boiler (71), events such as the operation of the ash removal equipment and the ignition and extinguishing of burners are determined by the main steam temperature, main steam pressure, reheat steam temperature, NOx concentration, etc. This is a disturbance for the feedback control of the amount (controlled amount), but the predicted fluctuation value (V ) Calculated based on

Specifically, using a function created based on control theory, experience, etc., a specified advance amount (Qa) is calculated according to, for example, the operating time of the ash removal device (the event duration (t)). Then, a leading indicator (Qb) is calculated based on the specified leading amount (Qa), the predicted fluctuation value (V), and the deviation (D) between the measured value (PV) of the controlled amount and the control set value (SV). The leading amount (Ff) is calculated by correcting it to a more appropriate value using . In other words, instead of directly calculating the preceding amount (Ff) based on the predicted fluctuation value (V) obtained from the learning model (M), the calculated value calculated based on this predicted fluctuation value (V) is The preceding amount (Ff) obtained immediately before is adjusted in a feedback control manner based on the result of applying it to the target operating end (9), and the specified preceding amount (Qa) is adjusted using the leading index (Qb) calculated in this way. ) is corrected to calculate the leading amount (Ff).

Even if the above-mentioned learning model (M) is used, there is a possibility that an error will occur between the predicted variation value (V) and the actual deviation (D), and leading indicators can only be obtained from the predicted variation value (V). When calculating (Qb), the predicted fluctuation value (V) is calculated with an error, but the leading indicator (Qb) is calculated based on the predicted fluctuation value (V) and the actual deviation (D). By calculating this, it is possible to obtain a leading indicator (Qb) that corresponds to a reduction in the error between the predicted fluctuation value (V) and the actual deviation (D). Therefore, it is possible to obtain a more appropriate preceding amount (Ff) according to the operating state of the plant (7) and its change over time, and by optimizing the manipulated variable (MV) of the target operating end (9), the control amount can be calculated. can be done more appropriately.

(2) In some embodiments, in the configuration of (1) above,
The second calculation unit (4)
a prescribed leading indicator calculation unit (42) that calculates a prescribed leading indicator value (Q1) according to the predicted fluctuation value (V);
an evaluation unit (43) that calculates an adjustment value (Q2) for adjusting the specified leading indicator value (Q1) based on the deviation (D);
and a leading indicator calculation unit (44) that provides the leading indicator (Qb) based on the prescribed leading indicator value (Q1) and the adjustment value (Q2).

According to configuration (2) above, by adjusting the uncorrected indicator calculated according to the predicted fluctuation value (V) with the adjustment value (Q2) calculated based on the deviation (D), the leading indicator (Qb) is calculated. This allows a more appropriate advance amount (Ff) to be determined.

(3) In some embodiments, in the configurations of (1) and (2) above,
The process parameter (P) includes a measured value of one or more state quantities, an operation amount of one or more operating terminals installed in the plant (7), and the leading indicator (Qb).

According to the configuration (3) above, the measurement value includes the measured value of the state quantity at the time of prediction, the operation amount of the operating end, and the leading indicator (Qb) (previous value). By including the measured value of the state quantity and the manipulated variable of the operating end, the operating state of the plant (7) can be appropriately specified by the process parameter (P). In addition, by including the leading indicator (Qb), the learning model (M) can output a predicted fluctuation value (V) that takes into account the leading indicator (Qb) at the previous specified timing, and the learning data collection time is reduced. The included learning time and the test run time of, for example, the boiler (71) for obtaining this learning data can be reduced.

(4) In some embodiments, in the configurations (1) to (3) above,
Further comprising a load acquisition unit (32) that acquires a load value (L) of a boiler (71) included in the plant (7),
The fluctuation prediction unit (2) calculates the load value based on the predicted fluctuation value (V) of each of the plurality of learning models (M) provided according to the plurality of load bands of the boiler (71). (L) is configured to output the predicted fluctuation value (V) according to
The advance amount calculation unit (3) includes the first calculation unit (31), the second calculation unit (4), and the third calculation unit (5), each of which is provided according to the plurality of load bands. However, it is configured to calculate the preceding amount (Ff) according to the load value (L).

According to configuration (4) above, the advance amount (Ff) is calculated in consideration of the load of the plant (7) (boiler (71)) at the time of calculation. With this, it is possible to obtain an appropriate advance amount (Ff) according to the load.

1 Control device 10 Control program 11 Processor 12 Storage unit 2 Fluctuation prediction unit 3 Advance amount calculation unit 31 First calculation unit 32 Load acquisition unit 4 Second calculation unit 41 Holding unit 42 Prescribed leading indicator calculation unit 43 Evaluation unit 43a Prescribed adjustment value Calculation unit 43b Output switching unit 43c Integrated value calculation unit 43d Adjustment value calculation unit 44 Leading indicator calculation unit 45 Switching unit 5 Third calculation unit 52 Switching unit 6 Feedback control unit 61 Deviation calculation unit 62 Controller 63 Addition unit 7 Plant 7L Cooling medium Supply pipe 7c Temperature reducer 7p Steam pipe 7s Pressure sensor 7t Boiler outlet temperature measurement sensor 7v Spray valve 7w Cooling medium 71 Boiler 71f Furnace 71p Furnace wall evaporation pipe 72 Combustion device 72v Fuel supply device 73 Flue 74 Combustion burner 75 Pulverized coal supply Pipe 76 Pulverizer 77a Air duct 77b Branch 77c Branch duct 77d Damper 77e Wind box 77p Port 78 Blower 79 Heat exchanger 79a Superheater 79b Reheater 79c Economizer 8 Exhaust gas passage 81 Denitration device 81p Ammonia supply pipe 81s NOx sensor 81v Ammonia flow control valve 8w Ammonia 82 Air heater 83 Dust treatment device 84 Induced blower 85 Chimney 86 Recirculation fan 87 Gas recirculation fan inlet damper 9 Target operating end T Steam turbine G Generator P Process parameter M Learning model V Predicted fluctuation value Ff Leading amount Fb First manipulated variable PV Measured value of controlled variable SV Control set value MV Manipulated amount D Deviation E Generated signal I Integrated value L Current load value Qa Specified leading amount Qb Leading indicator Q1 Specified leading indicator value Q2 Adjustment value Q3 Specified adjustment value t duration of event

Claims (4)

A control device that determines the operation amount of a target operating end that can control the control amount of a plant at each prescribed timing,
Outputting a predicted variation value of the controlled variable according to the input of the process parameter using a learning model that has learned the relationship between the process parameter for identifying the operating state of the plant and the variation value of the controlled variable. a fluctuation prediction unit configured as such,
a preceding amount calculation unit configured to calculate, based on the predicted fluctuation value, a preceding amount for correcting a first manipulated variable calculated according to a deviation between a measured value of the controlled variable and a control setting value; , comprising:
The advance amount calculation unit is
a first calculation unit configured to calculate, when an event that changes the control amount, a specified preceding amount according to a duration of the event;
a second calculation unit configured to calculate a leading indicator for correcting the prescribed leading amount based on the predicted fluctuation value and the deviation;
A control device comprising: a third calculation unit configured to calculate the lead amount based on the prescribed lead amount and the lead indicator. The second calculation unit is
a prescribed leading indicator calculation unit that calculates a prescribed leading indicator value according to the predicted fluctuation value;
an evaluation unit that calculates an adjustment value for adjusting the prescribed leading indicator value based on the deviation;
The control device according to claim 1, further comprising a leading indicator calculation unit that provides the leading indicator based on the prescribed leading indicator value and the adjustment value. The control device according to claim 1 or 2, wherein the process parameter includes a measured value of one or more state quantities, a manipulated variable of one or more operating terminals installed in the plant, and the leading indicator. further comprising a load acquisition unit that acquires a load value of a boiler included in the plant,
The fluctuation prediction unit is configured to output the predicted fluctuation value according to the load value based on the predicted fluctuation value of each of the plurality of learning models provided according to the plurality of load bands of the boiler. It is configured,
The advance amount calculation section includes the first calculation section, the second calculation section, and the third calculation section, each of which is provided according to the plurality of load bands, and calculates the advance amount according to the load value. The control device according to any one of claims 1 to 3, wherein the control device is configured to calculate.

Images