JP2023175210A - Control apparatus for wet flue-gas desulfurization device, remote monitoring system, and control method - Google Patents

Abstract

The present invention provides a control device in which the accuracy of a table for determining a control target value for a wet flue gas desulfurization device is improved. A control device 15 for a wet flue gas desulfurization device for combustion device exhaust gas, comprising a learning model construction section, a predicted value correction section, a table creation section, and a control target value determination section, the learning model construction section comprising: A learning model is constructed for the relationship between explanatory variables including the load of the combustion equipment and the sulfur dioxide concentration at the outlet of the absorption tower, which is the objective variable, and the predicted value correction section calculates the predicted value by the learning model based on the difference from the actual measured value. The table creation unit creates a table showing the relationship between the load of the combustion device and the absorbent concentration target value and absorbent circulation amount target value so that the corrected predicted value satisfies the reference value, and calculates the control target. The value determination unit calculates the absorbent concentration target value based on the table and determines the control target value for the amount of absorbent input and the circulation flow rate of the absorbent, and the predicted value correction unit calculates the predicted value based on the load. This is a control device including a first correction section that performs correction. [Selection diagram] Figure 2

Description

The present disclosure relates to a control device, a remote monitoring system, and a control method for a wet flue gas desulfurization device.

In a wet flue gas desulfurization device, exhaust gas generated in a combustion device such as a boiler is introduced into an absorption tower of the desulfurization device, and brought into gas-liquid contact with an absorption liquid circulating through the absorption tower. During the gas-liquid contact process, the absorbent (e.g., calcium carbonate) in the absorption liquid reacts with sulfur dioxide (SO ₂ ) in the exhaust gas, so that SO ₂ in the exhaust gas is absorbed by the absorption liquid and removed from the exhaust gas. SO ₂ is removed (exhaust gas is desulfurized). On the other hand, the absorption liquid that has absorbed SO ₂ falls and is stored in a storage tank below the absorption tower. An absorbent is supplied to the storage tank, and the absorbent whose absorption performance has been restored by the supplied absorbent is supplied to the upper part of the absorption tower by a circulation pump, where it is brought into gas-liquid contact with the exhaust gas (absorption of SO ₂ ). It will be done. Since the circulation pump that circulates the absorption liquid consumes a large amount of power, conventionally, in order to reduce power consumption, the amount of absorption liquid required is determined based on the flow rate of the exhaust gas flowing into the absorption tower, the SO ₂ concentration in the exhaust gas, etc. The circulation flow rate is calculated and the number of operating circulation pumps is controlled.

Patent Document 1 discloses a technique for appropriately adjusting the operating conditions of a circulation pump for circulating an absorption liquid in an absorption tower of such a wet flue gas desulfurization apparatus. This document uses operational data obtained from combustion equipment such as boilers and wet flue gas desulfurization equipment to examine the correlation between the operational data and the SO ₂ concentration at the outlet of the absorption tower, and the correlation between the operational data and the SO 2 concentration contained in the absorption liquid. The correlation with the absorbent concentration is modeled by machine learning, and the circulation flow rate of the absorbent and the absorbent concentration are controlled to be optimized based on the tables obtained by these two learning models.

JP 2020-11163 Publication

In Patent Document 1, the control target values for the circulating flow rate of the absorption liquid and the concentration of the absorbent are calculated based on the table determined by the learning model, so if the prediction accuracy of the learning model decreases due to some factor (the learning model If the predicted value deviates from the actual measured value), the control target value calculated based on the table will deviate from the optimal value, and there is a risk that control accuracy will decrease. In order to suppress such a decrease in control accuracy, when the prediction accuracy of the learning model decreases, it may be possible to improve the learning model by rebuilding it through relearning. However, relearning of a learning model requires high processing power from a processing device such as a computer that performs calculations, and the cost to realize it becomes high. Another method to improve the prediction accuracy of the learning model is to perform correction processing on the predicted values of the learning model without retraining the learning model. The accuracy of the predicted value depends on the condition of the wet flue gas desulfurization equipment. Therefore, even when improving prediction accuracy by correcting the predicted value of the learning model, the problem is how to perform the correction process.

At least one embodiment of the present disclosure has been made in view of the above circumstances, and provides a wet flue gas desulfurization device that can improve the accuracy of a table for determining control target values by correcting predicted values of a learning model. The purpose of the present invention is to provide a control device, a remote monitoring system, and a control method.

In order to solve the above problems, a control device for a wet flue gas desulfurization device according to at least one embodiment of the present disclosure,
A control device for a wet flue gas desulfurization device that performs desulfurization by bringing exhaust gas generated in a combustion device into gas-liquid contact with an absorption liquid in an absorption tower,
At least one parameter included in the operation data of at least one of the combustion device or the wet flue gas desulfurization device, the explanatory variable including at least the load of the combustion device, and the future sulfur dioxide concentration at the outlet of the absorption tower. a learning model construction unit for constructing a learning model using machine learning regarding the relationship with the objective variable;
Predicted value correction for correcting the predicted value of the sulfur dioxide concentration by the learning model using a correction coefficient calculated based on the difference between the predicted value and the actual measured value of the sulfur dioxide concentration at the outlet of the absorption tower. Department and
A table showing the relationship between the load of the combustion device and the absorbent concentration target value and absorbent circulation amount target value of the absorbent liquid so that the corrected predicted value corrected by the predicted value correction unit satisfies the reference value. A table creation part for creating,
The absorbent concentration target value and the absorbent circulation rate target value corresponding to the operation data are calculated based on the table, and the absorbent concentration target value and the absorbent circulation rate target value corresponding to the absorbent concentration target value and the absorbent circulation rate target value are calculated. a control target value determination unit for determining a control target value of the input amount and the circulating flow rate of the absorption liquid;
Equipped with
The predicted value correction unit includes a first correction unit that corrects the predicted value using a first correction coefficient calculated based on the load as the correction coefficient.

In order to solve the above problems, a method for controlling a wet flue gas desulfurization device according to at least one embodiment of the present disclosure includes:
A method for controlling a wet flue gas desulfurization device that performs desulfurization by bringing exhaust gas generated in a combustion device into gas-liquid contact with an absorption liquid in an absorption tower, the method comprising:
At least one parameter included in the operation data of at least one of the combustion device or the wet flue gas desulfurization device, the explanatory variable including at least the load of the combustion device, and the future sulfur dioxide concentration at the outlet of the absorption tower. A step of constructing a learning model using machine learning regarding the relationship with the objective variable;
correcting the predicted value of the sulfur dioxide concentration by the learning model using a correction coefficient calculated based on the difference between the predicted value and the actual measured value of the sulfur dioxide concentration at the outlet of the absorption tower;
A table showing the relationship between the load of the combustion device and the absorbent concentration target value and absorbent circulation amount target value of the absorbent liquid so that the corrected predicted value corrected by the predicted value correction unit satisfies the reference value. The process of creating
The absorbent concentration target value and the absorbent circulation rate target value corresponding to the operation data are calculated based on the table, and the absorbent concentration target value and the absorbent circulation rate target value corresponding to the absorbent concentration target value and the absorbent circulation rate target value are calculated. determining control target values for the input amount and the circulating flow rate of the absorption liquid;
Equipped with
In the step of correcting the predicted value, the predicted value is corrected using a first correction coefficient calculated based on the load as the correction coefficient.

According to at least one embodiment of the present disclosure, a control device and a remote monitoring system for a wet flue gas desulfurization device that can improve the accuracy of a table for determining a control target value by correcting a predicted value of a learning model; Also, a control method can be provided.

FIG. 1 is a configuration diagram of a wet flue gas desulfurization device according to an embodiment. FIG. 2 is a block configuration diagram of the control device in FIG. 1. FIG. FIG. 3 is a processing flow diagram of the necessary condition determination unit in FIG. 2; FIG. 3 is a processing flow diagram of the first correction condition determining section in FIG. 2; FIG. 3 is a processing flow diagram of the load-specific correction signal generation section of FIG. 2; FIG. 3 is a processing flow diagram of a first load-specific correction coefficient calculation unit among the load-specific correction coefficient calculation units in FIG. 2; FIG. 3 is a processing flow diagram of a first correction coefficient calculation unit in FIG. 2; FIG. 3 is a processing flow diagram of a second correction condition determining section in FIG. 2; FIG. 3 is a processing flow diagram of a second correction coefficient calculation unit in FIG. 2; FIG. 3 is a processing flow diagram of a table creation unit in FIG. 2; FIG. 3 is a process flow diagram regarding correction of the control target value by the control target value determination unit of FIG. 2; FIG. 3 is a processing flow diagram of the circulation pump adjustment section of FIG. 2. FIG. 3 is a block configuration diagram showing another aspect of FIG. 2. FIG. 3 is a block configuration diagram showing another aspect of FIG. 2. FIG.

Hereinafter, some embodiments of the present invention will be described with reference to the drawings. However, the scope of the present invention is not limited to the following embodiments. The dimensions, materials, shapes, relative positions, etc. of the structures described in the following embodiments are not intended to limit the scope of the present invention, but are merely illustrative examples.

FIG. 1 is a configuration diagram of a wet flue gas desulfurization apparatus 10 according to an embodiment.
The wet flue gas desulfurization device 10 is a device for desulfurizing the exhaust gas generated by the combustion device 1. The combustion device 1 is, for example, a boiler for generating steam. The steam generated by the combustion device 1 is supplied to, for example, a steam turbine (not shown), and when the steam turbine is driven by the steam, electric power is generated by the generator 5 connected to the output shaft of the steam turbine. The wet flue gas desulfurization device 10 includes an absorption tower 11 that communicates with the combustion device 1 via a pipe 2, and a plurality of circulation pumps 12a, 12b, which are provided in a pipe 3 for circulating an absorption liquid circulating inside the absorption tower 11. 12c,... (In FIG. 1, three circulation pumps are representatively illustrated, and the number is not limited. In addition, when these are collectively referred to as "circulation pump 12"), An absorbent slurry supply section 13 for supplying slurry (absorbent slurry) of calcium carbonate (CaCO ₃ ), which is an absorbent contained in the liquid, into the absorption tower 11, and a gypsum for recovering gypsum in the absorption liquid. The recovery section 14 is also provided. The absorption tower 11 is provided with an outflow pipe 16 for the exhaust gas desulfurized in the operation described below to flow out from the absorption tower 11 as an outflow gas, and the outflow pipe 16 is provided with a pipe for measuring the SO ₂ concentration in the outflow gas. A gas analyzer 17 is provided.

The absorbent slurry supply unit 13 includes an absorbent slurry production facility 21 for producing absorbent slurry, an absorbent slurry supply pipe 22 that communicates the absorbent slurry production facility 21 and the absorption tower 11, and an absorbent slurry supply pipe 22 for communicating the absorbent slurry production facility 21 with the absorption tower 11. It includes an absorbent slurry supply amount control valve 23 for controlling the flow rate of the absorbent slurry flowing through the supply piping 22. The gypsum recovery unit 14 includes a gypsum separator 25, a gypsum slurry extraction pipe 26 that communicates the gypsum separator 25 and the absorption tower 11, and a gypsum slurry extraction pump 27 provided in the gypsum slurry extraction pipe 26. We are prepared.

The wet flue gas desulfurization device 10 is provided with a control device 15 for controlling the wet flue gas desulfurization device 10. The control device 15 acquires various operational data (for example, temperatures and pressures at various parts, flow rates of various fluids, etc.) of the combustion device 1 and the wet flue gas desulfurization device 10, and performs various controls.

The hardware configuration of the control device 15 includes, for example, a CPU (Central Processing Unit), a RAM (Random Access Memory), a ROM (Read Only Memory), a computer-readable storage medium, and the like. A series of processes for realizing various functions is stored in a storage medium, etc. in the form of a program, for example, and the CPU reads this program into a RAM, etc., and executes information processing and arithmetic processing. By doing so, various functions are realized. Note that the program may be pre-installed in a ROM or other storage medium, provided as being stored in a computer-readable storage medium, or distributed via wired or wireless communication means. etc. may also be applied. Computer-readable storage media include magnetic disks, magneto-optical disks, CD-ROMs, DVD-ROMs, semiconductor memories, and the like.

FIG. 2 is a block diagram of the control device 15 in FIG. 1. In this example, the control device 15 is configured to include an upper control device 15A and a lower control device 15B. The upper control device 15A is a main device that constitutes the control device 15, and is realized by having each functional block described by a program. The lower control device 15B is a control device placed under the control of the upper control device 15A, and is, for example, a distributed control system (DCS) of each device that constitutes the wet flue gas desulfurization device 10 (see FIG. 1). This is achieved by describing each functional block using control logic. These upper control device 15A and lower control device 15B can communicate with each other and control the wet flue gas desulfurization device 10 in cooperation with each other.

The control device 15 includes an operation data reception section 30, a learning model construction section 38, a table creation section 31, a predicted value correction section 36, a control target value determination section 32, a circulation pump adjustment section 33, and an absorbent slurry supply control section 34. Equipped with Each of the functional blocks of the control device 15 is capable of transmitting and receiving data to and from each other, and can cooperate to realize the control method described below. In this embodiment, among these functional blocks, the driving data receiving section 30, the learning model constructing section 38, and the table creating section 31 are arranged in the upper control device 15A, while the predicted value correcting section 36, the control target value determining section 32, the circulation pump adjustment section 33, and the absorbent slurry supply control section 34 are arranged in the lower control device 15B.

Note that the control device 15 includes a relearning device 15C in addition to the upper control device 15A and the lower control device 15B. The relearning device 15C improves the prediction accuracy of the learning model M by relearning the learning model M based on a relearning algorithm when the prediction accuracy of the learning model M constructed by the learning model construction unit 38 decreases. This is a configuration for improving. If the prediction accuracy of the learning model M decreases, it can be corrected by correcting the predicted value Vp by the predicted value correction unit 36 as described later, but if the prediction accuracy cannot be corrected by correcting the predicted value Vp by the predicted value correction unit 36, as described later. In this case, the learning model M can be relearned by the relearning device 15C. Since such relearning is not performed frequently, the processing load on the upper controller 15A and lower controller 15B can be reduced by providing the relearning device 15C as a separate configuration from the upper controller 15A and lower controller 15B. can be effectively reduced.

The driving data receiving section 30 is configured to receive the driving data acquired by the driving data acquiring section 20. This operation data includes various parameters acquired by the combustion device 1 and the wet flue gas desulfurization device 10, and in particular, the load of the combustion device 1 and the SO ₂ at the outlet of the absorption tower measured by the gas analyzer 17. Concentration is included at least.

The learning model constructing unit 38 is configured to construct a learning model M by machine learning regarding the relationship between various operating data received by the operating data receiving unit 30 and future SO ₂ concentration at the outlet of the absorption tower. The learning model M is constructed as a regression model using a regression method such as multiple regression, ridge regression, lasso regression, or elastic net. In this embodiment, as an example of the learning model M, a regression model expressed by a linear polynomial as shown in the following equation is constructed.
SO ₂ concentration at the absorption tower outlet = k1 x explanatory variable 1 + k2 x explanatory variable 2 + ... + kn x explanatory variable n + b (1)
By using a linear polynomial as the learning model M in this way, the explainability (interpretability) is higher than that of a complicated simulation model, and the calculation load can be effectively reduced. Note that n is an arbitrary natural number, k1 to kn are coefficients, and b is an arbitrary intercept.

The learning model M obtained by machine learning uses an explanatory variable consisting of a plurality of parameters included in the operation data received by the operation data receiving unit 30 and the future SO ₂ concentration at the absorption tower outlet as an objective variable, and calculates the correlation between the two. It is constructed as a model to show the following. Here, the combination of a plurality of parameters included in the explanatory variables of the learning model M can be arbitrarily selected from the following candidates.
i) Output command value for generator 5 (output command value from outside)
ii) Output of the generator 5 iii) Supply air flow rate to the combustion device 1 or exhaust gas flow rate from the combustion device 1 iv) SO ₂ concentration at the absorption tower 11 inlet or SO ₂ concentration at the combustion device 1 outlet v) at the absorption tower 11 exit _SO2 concentration vi) _CaCO3 concentration or pH of absorption liquid
vii) Control value of the number of operating circulation pumps 12 or discharge flow rate viii) O ₂ concentration at the exit of the combustion device 1 or the inlet of the absorption tower 11 Sulfur concentration/sulfur concentration at the inlet of absorption tower 11) x 100%)
These candidates are parameters that can be conventionally measured in many wet flue gas desulfurization apparatuses 10, and may include other parameters.

In this embodiment, the explanatory variables of the learning model M are selected from among the above candidates: i) output command value for the generator 5 (output command value from the outside), iii) supply air flow rate for the combustion device 1 or It is selected to include at least one of the exhaust gas flow rate, or iv) the SO ₂ concentration at the absorption tower 11 inlet or the SO ₂ concentration at the combustion device 1 outlet. More preferably, the explanatory variables of the learning model are, from among the above candidates, iii) the supply air flow rate to the combustion device 1 or the exhaust gas flow rate from the combustion device 1, and iv) the SO ₂ concentration at the inlet of the absorption tower 11 or the combustion device 1 chosen to include the _SO2 concentration at the outlet. Thereby, it is possible to construct a learning model M that can predict the future SO ₂ concentration at the outlet of the absorption tower, which is the objective function, with good accuracy. In addition, by selecting some parameters as explanatory variables from the driving data received by the driving data receiving unit 30, it is possible to efficiently narrow down the learning target data and reduce the computational burden of machine learning. .

The table creation unit 31 uses the learning model M constructed by the learning model construction unit 38 to calculate the output of the generator 5 (load of the combustion device 1) and the predicted value Vp of the future SO ₂ concentration at the outlet of the absorption tower. This configuration is for creating a table Tb that defines the relationship between the absorption liquid and the absorbent concentration target value to satisfy the reference value. A specific method for creating the table Tb by the table creation unit 31 will be described later, but the accuracy of the table Tb created by the table creation unit 31 depends on the accuracy of the predicted value Vp calculated based on the learning model M. do. Therefore, when the prediction accuracy of the learning model M is low (that is, when there is a discrepancy between the predicted value Vp and the actual value Vm), the predicted value correction unit 36 corrects the predicted value Vp of the learning model M. The table creation unit 31 can improve the calculation accuracy of the control target value by creating the table Tb using the predicted value Vp corrected in this way (hereinafter appropriately referred to as "corrected predicted value Vp'"). .

The control target value determination unit 32 determines the absorbent concentration target value corresponding to the load included in the operation data received by the operation data reception unit 30 based on the table Tb created by the table creation unit 31, and This configuration is for determining respective control target values of the amount of absorbent input and the circulating flow rate of the absorbent liquid corresponding to the target agent concentration value.

Further, the circulation pump adjustment section 33 is configured to control the circulation pumps 12a to 12c based on the control target value regarding the circulation flow rate of the absorption liquid determined by the control target value determination section 32. Further, the absorbent slurry supply control section 34 is configured to control the supply amount of the absorbent based on the control target value regarding the input amount of the absorbent determined by the control target value determination section 32.

Next, the configuration of the predicted value correction section 36 will be explained in detail. The predicted value correction unit 36 includes a necessary condition determination unit 39 for determining whether correction processing for the predicted value Vp of the learning model M is necessary. The necessary condition determining unit 39 is configured to determine whether or not the predicted value Vp of the learning model M constructed by the learning model constructing unit 38 requires correction. The learning model M includes a considerable amount of prediction error, and if the difference ΔV between the predicted value Vp and the actual measurement value Vm becomes large, the learning model M will be corrected by at least one of the first correction unit 40 and the second correction unit 50, which will be described later. The predicted value Vp by the model M is corrected. The necessary conditions are defined as conditions for determining whether correction processing is necessary for the predicted value Vp of such learning model M.

FIG. 3 is a processing flow diagram of the necessary condition determination unit 39 in FIG. 2. FIG. 3 shows, as an example of the configuration of the necessary condition determination unit 39, a processing flow when determination is made based on necessary conditions consisting of the following seven conditions. Specifically, the necessary condition determining unit 39 determines that the necessary condition is satisfied when all of the following seven conditions (conditions 1 to 7) are satisfied.
(Condition 1) The measured value Vm (SO ₂ concentration at the absorption tower outlet) corresponding to the predicted value Vp of the learning model M is larger than a preset reference value Vmref.
(Condition 2) The load L of the combustion device 1 is greater than a preset reference value Lref.
(Condition 3) There is no abnormality in the parameters that are correlated with the explanatory variables of the learning model M.
(Condition 4) The bypass damper is in a closed state.
(Condition 5) The wet flue gas desulfurization device 10 is in a state where smoke is flowing.
(Condition 6) There is no abnormality in the host controller 15A.
(Condition 7) Initialization of the control device 15 is completed.

Condition 1 is whether or not it is necessary to correct the predicted value Vp of the learning model M based on whether the actual measured value Vm of the SO ₂ concentration at the absorption tower outlet, which is the prediction target of the learning model M, is greater than the reference value Vref. This is the condition for determining. Condition 2 is a condition for determining whether it is necessary to correct the predicted value Vp of the learning model M based on whether the load L of the combustion device 1 is larger than the reference load value Lref.
Note that since Condition 1 and Condition 2 are conditions for determining whether or not correction is necessary from a similar viewpoint, either one may be omitted.

Condition 3 is a condition for determining whether the vehicle is in a driving state suitable for correcting the predicted value Vp of the learning model M. The "parameters that are correlated with the explanatory variables of the learning model M" may be, for example, parameters included in the explanatory variables (SO ₂ concentration at the outlet of the absorption tower, exhaust gas flow rate of the combustion device 1, absorbent concentration, etc.). , or may be a parameter for monitoring abnormalities in a measuring device for measuring these parameters.

Condition 4 is a condition for determining whether the bypass damper included in the wet flue gas desulfurization device 10 is in the closed state. A bypass damper is a damper valve that bypasses the boiler exhaust gas directly to the chimney without passing it through the desulfurization equipment.It is closed when the power plant is in normal operation, but it is closed when some abnormality occurs, for example. The structure is designed to protect the desulfurization equipment by fully opening it in case of an emergency. Incidentally, in recent years, some plants are not equipped with a bypass damper, but in that case, condition 4 can be omitted from the necessary conditions.

Condition 5 is a condition for determining whether or not the wet type flue gas desulfurization device 10 is in a smoking state, for example, whether or not the desulfurization fan included in the wet type flue gas desulfurization device 10 is in operation, or The determination is made based on whether or not the IDF is activated.

Condition 6 is a condition for determining whether there is any abnormality in the upper control device 15A of the control device 15, and condition 7 is a condition for determining whether initialization of the control device 15 has been completed. This is the condition.

When all of these conditions are met, the necessary condition determining unit 39 determines that the necessary conditions are satisfied and turns on the necessary condition fulfillment flag. When the necessary condition fulfillment flag is turned ON, the predicted value Vp can be corrected by the first correction unit 40 and the second correction unit 50.

The first correction unit 40 is configured to correct the predicted value Vp based on the learning model M in accordance with the load L of the combustion device 1. The first correction section 40 includes a first correction condition determination section 41 , a load-specific correction signal generation section 42 , a load-specific correction coefficient calculation section 43 , and a first correction coefficient calculation section 44 . Each component included in the first correction section 40 will be described in order below with reference to FIGS. 4 to 7.

FIG. 4 is a processing flow diagram of the first correction condition determination unit 41 in FIG. 2. The first correction condition determination unit 41 is a configuration for determining whether the first correction condition is satisfied or not. The first correction condition is a condition for determining whether or not to perform the first correction by the first correction unit 40 on the predicted value Vp of the learning model M. In FIG. 4, an example of the first correction condition is shown. Judgments are made based on the following four conditions.
(Condition 8) The difference ΔV between the corrected predicted value Vp′ (predicted value Vp before the first correction) and the actual measured value Vm is larger than the reference value ΔVref.
(Condition 9) The necessary condition fulfillment flag is ON.
(Condition 10) The first corrected frequency signal Sa1 is input.
(Condition 11) The first correction permission flag is ON.

In condition 8, the absolute value (ABS) is calculated for the difference ΔV between the corrected predicted value Vp' (predicted value Vp of the learning model M itself before the first correction) and the actual measured value Vm, and the absolute value is calculated from the reference value ΔVref. It is determined whether or not it is large. In FIG. 4, the measured value Vm that is compared with the corrected predicted value Vp' of the learning model M is corrected for the measurement delay. The measurement delay correction is performed to correct the time lag (measurement delay) of the actual measurement value Vm by the gas analyzer 17 with respect to the corrected predicted value Vp', since acquisition of the actual measurement value Vm by the gas analyzer 17 requires a considerable amount of measurement time. It is processing.

Condition 9 is a condition for determining whether the necessary condition fulfillment flag, which can be switched by the output of the above-mentioned necessary condition determination unit 39, is ON. By including condition 9 in the first correction conditions in this way, the first correction by the first correction unit 40 is performed on the premise that the above-mentioned necessary conditions are satisfied.

Condition 10 is a condition for adjusting the frequency of execution by the first correction unit 40 based on the input first correction frequency signal Sa1. The first correction frequency signal Sa1 is a rectangular wave pulse signal that repeats an ON time for validating the first correction and an OFF time for invalidating the first correction. When the first corrected frequency signal Sa1 is input, condition 10 is determined to be satisfied during the ON time, and determined not to be satisfied during the OFF time. In condition 10, the frequency of performing the first correction can be adjusted by changing the ratio between the ON time and the OFF time of the input first correction frequency signal Sa1. The ON time and OFF time of the first corrected frequency signal Sa1 can be set by the user.

Condition 11 is a condition for determining whether the first correction permission flag for determining permission to implement the first correction is ON. For example, the control device 15 includes an operation button (not shown) that allows the user to select whether or not to perform the first correction, and when the operation button is turned on, the first correction permission flag is turned on. This allows the user to select whether or not to perform the first correction.

The first correction condition determination unit 41 determines that the first correction condition is satisfied when all of these conditions 8 to 11 are satisfied, and outputs the first correction signal S1. As described above, the first correction frequency signal Sa1, which is a rectangular wave pulse signal, is input to the first correction condition as condition 10. Therefore, the first correction signal S1 is adjusted so as to correspond to the first correction frequency signal Sa1. It is output as a rectangular wave pulse signal (flicker signal) in which ON time and OFF time are repeated.

The load-specific correction signal generation unit 42 is configured to generate a load-specific correction signal Sl according to the load L of the combustion device 1. Here, FIG. 5 is a processing flow diagram of the load-specific correction signal generation section 42 of FIG. 2. In FIG.

In this embodiment, the load range that the combustion device 1 can take is divided into a first load range Lr1 to a fourth load range Lr4 (Lr1<Lr2<Lr3<Lr4), and the load-specific correction signal generation unit 42 The first load range determining section 42a1 to the fourth load range determining section 42a4 corresponding to the load range determining section 42a4 are provided. The first load range determination unit 42a1 is a logic circuit that outputs an ON signal when the input load L is included in the first load range Lr1, and outputs an OFF signal when it is not included in the first load range Lr1. be. The second load range determination unit 42a2 is a logic circuit that outputs an ON signal when the input load L is included in the second load range Lr2, and outputs an OFF signal when it is not included in the second load range Lr2. be. The third load range determination unit 42a3 is a logic circuit that outputs an ON signal when the input load L is included in the third load range Lr3, and outputs an OFF signal when it is not included in the third load range Lr3. be. The fourth load range determination unit 42a4 is a logic circuit that outputs an ON signal when the input load L is included in the fourth load range Lr4, and outputs an OFF signal when it is not included in the fourth load range Lr4. be.

The first load-specific correction signal output section 42b1 is configured to output an ON signal from the first load range determination section 42a1, and to which the first correction signal S1 (flicker signal) from the first correction condition determination section 41 described above is in an ON state. In this case, a first load-specific correction signal Sl1 is output for instructing to perform correction corresponding to the first load range Lr1. The second load-based correction signal output unit 42b2 is configured to output an ON signal from the second load range determination unit 42a2, and to which the first correction signal S1 (flicker signal) from the first correction condition determination unit 41 is in the ON state. In this case, a second load-specific correction signal Sl2 is output for instructing to perform correction corresponding to the second load range Lr2. The third load-specific correction signal output unit 42b3 is configured to output an ON signal from the third load range determination unit 42a3, and to which the first correction signal S1 (flicker signal) from the first correction condition determination unit 41 is in the ON state. In this case, a third load-specific correction signal Sl3 is output for instructing to perform correction corresponding to the third load range Lr3. The fourth load-based correction signal output unit 42b4 is configured to output an ON signal from the fourth load range determination unit 41a4, and to which the first correction signal S1 (flicker signal) from the first correction condition determination unit 41 is in the ON state. In this case, a fourth load-specific correction signal Sl4 is output for instructing to perform correction corresponding to the fourth load range Lr4.

The load-specific correction coefficient calculation unit 43 is configured to calculate load-specific correction coefficients Y1 to Y4 corresponding to the load-specific correction signals Sl1 to Sl4 generated by the load-specific correction signal generation unit 42. Here, FIG. 6 is a processing flow diagram of the first load-based correction coefficient calculation unit 43a of the load-based correction coefficient calculation unit 43 in FIG. The first load-specific correction coefficient calculation section 43a is a configuration of the load-specific correction coefficient calculation section 43 for calculating the first load-specific correction coefficient Y1 based on the first load-specific correction signal Sl1.

In addition to the first load-specific correction coefficient calculation unit 43a, the load-specific correction coefficient calculation unit 43 includes a second load-specific correction coefficient calculation unit 43 for calculating the second load-specific correction coefficient Y2 based on the second load-specific correction signal Sl2. A correction coefficient calculation unit 43b, a third load-specific correction coefficient calculation unit 43c for calculating a third load-specific correction coefficient Y3 based on the third load-specific correction signal Sl3, and a fourth load-specific correction coefficient calculation unit 43c based on the fourth load-specific correction signal Sl4. A fourth load-specific correction coefficient calculation unit 43d is provided for calculating the fourth load-specific correction coefficient Y4. Although not shown, the second load-specific correction coefficient calculation unit 43b, the third load-specific correction coefficient calculation unit 43c, and the fourth load-specific correction coefficient calculation unit 43d are described below unless otherwise specified. This is the same as the per-load correction coefficient calculation unit 43a.

In the first load-specific correction coefficient calculation unit 43a, when the difference ΔV between the corrected predicted value Vp' (the predicted value Vp of the learning model M before the first correction) and the actual measured value Vm is input, the difference ΔV is negative. In some cases (that is, when the predicted value Vp is smaller than the actual value Vm), the first switch T1 is switched to select the upward correction value A1. The upward correction value A1 is set as a coefficient of 1 or more obtained by adding the upward correction width to the reference value "1", and is, for example, "1.01". The upward correction value A1 output from the first switch T1 is multiplied by a correction gain K (typically set to "1.0", but can be changed as appropriate) set in advance by the correction gain adjustment section P. After that, the first correction signal S1 output from the first correction condition determining section 41 described above is output to the second switch T2 at the timing when the first correction signal S1 is turned on (note that the first correction signal S1 input to the second switch T2 At the timing when the correction signal S1 is in the OFF state, the second switch T2 outputs the default value "1"). The output of the second switch T2 is multiplied by the previous value (first load-specific correction coefficient) stored in advance, and is output as the first load-specific correction coefficient Y1. The first load-based correction coefficient Y1 output in this way is stored in a storage unit such as a memory (not shown), and is used as the previous value in the next calculation cycle.

In this way, when the difference ΔV is negative, the first load-specific correction coefficient Y1 increases every calculation cycle by the upward correction width set in the upward correction value A1. As a result, when the corrected predicted value Vp' is smaller than the actual measured value Vm, the first load-specific correction coefficient Y1 is calculated so that the corrected predicted value Vp' is corrected in the direction of increasing and approaches the actual measured value Vm. Ru.

On the other hand, when the difference ΔV is positive (that is, when the corrected predicted value Vp' is larger than the actual measured value Vm), the first switch T1 is switched to select the downward correction value A2. The downward correction value A2 can be set as a coefficient less than 1 obtained by adding a downward correction width (negative value) to the reference value "1", and is, for example, "0.99". The downward correction value A2 output from the first switch T1 is multiplied by the correction gain K (typically set to "1", but can be changed as appropriate) set in advance by the correction gain adjustment section P. , is outputted to the second switch T2 at the timing when the first correction signal S1 outputted from the first correction condition determining section 41 described above turns ON (the first correction signal inputted to the second switch T2 At the timing when T2 becomes OFF, the second switch T2 outputs the default value "1"). The output of the second switch T2 is multiplied by the previous value (first load-specific correction coefficient) stored in advance, and is output as the first load-specific correction coefficient Y1. The first load-based correction coefficient Y1 output in this way is stored in a storage unit such as a memory (not shown), and is used as the previous value in the next calculation cycle.

In this way, when the difference ΔV is positive, the first load-based correction coefficient Y1 decreases every calculation cycle by the downward correction width set in the downward correction value A2. As a result, when the corrected predicted value Vp' is larger than the actual measured value Vm, the first load-specific correction coefficient Y1 is calculated so that the corrected predicted value Vp' is corrected in a decreasing direction and approaches the actual measured value Vm. Ru.

The first correction coefficient calculation unit 44 is configured to correct the predicted value Vp by the learning model M using the first correction coefficient Y corresponding to the load L. Here, FIG. 7 is a processing flow diagram of the first correction coefficient calculation unit 44 of FIG. 2.

The first correction coefficient calculation unit 44 receives the first load correction coefficient Y1 to fourth load correction coefficient Y4 for each load range calculated by the load correction coefficient calculation unit 43. Then, when the load L included in the operation data is input, one of the first load correction coefficient Y1 to the fourth load correction coefficient Y4 corresponding to the load range to which the load L belongs is set as the first correction coefficient Y. Calculated. The first correction coefficient Y calculated in this way has a value corresponding to the magnitude of the load L.

The first correction coefficient Y calculated by the first correction coefficient calculation unit 44 is output from the third switch T3 and multiplied by the predicted value Vp of the learning model M when the first correction permission flag is turned ON. By doing so, a corrected predicted value Vp' that has been subjected to the first correction is obtained. Such a first correction is performed by multiplying the predicted value Vp by a first correction coefficient Y calculated according to which of the first load range Lr1 to fourth load range Lr4 the load L belongs to. be exposed. Therefore, the amount of correction for the predicted value Vp can be finely set depending on the value of the load L, and the prediction accuracy can be effectively improved without reconstructing the learning model M.

The second correction unit 50 corrects the predicted value Vp by the learning model M according to the total load L of the combustion device 1 (that is, unlike the first correction that depends on the load L, This is a configuration for performing corrections that do not depend on The second correction section 50 includes a second correction condition determination section 51 and a second correction coefficient calculation section 52.

FIG. 8 is a processing flow diagram of the second correction condition determining section 51 of FIG. 2. The second correction condition determination unit 51 is configured to determine whether the second correction condition is satisfied. The second correction condition is a condition for determining whether or not to perform the second correction by the second correction unit 50 on the predicted value Vp of the learning model M. In FIG. 8, an example of the second correction condition is Judgments are made based on the following four conditions.
(Condition 12) The difference ΔV between the result of multiplying the corrected predicted value Vp' (predicted value Vp before the first correction) by the overall adjustment coefficient and the actual measured value Vm is larger than the reference value ΔVref.
(Condition 13) The necessary condition fulfillment flag is ON.
(Condition 14) The second corrected frequency signal Sa2 is input.
(Condition 15) The second correction permission flag is ON.

In condition 12, the absolute value (ABS) of the difference ΔV between the result of multiplying the corrected predicted value Vp' (predicted value Vp of the learning model M itself before the first correction) by the overall adjustment coefficient and the actual measured value Vm is It is determined whether the absolute value is larger than the reference value ΔVref. In FIG. 8, the measured value Vm that is compared with the corrected predicted value Vp' of the learning model M is corrected for the measurement delay. The measurement delay correction is performed to correct the time lag (measurement time delay) between the actual measurement value Vm by the gas analyzer 17 and the corrected predicted value Vp', since acquisition of the actual measurement value Vm by the gas analyzer 17 requires considerable measurement time. This is the process.

Condition 13 is a condition for determining whether the necessary condition fulfillment flag, which can be switched by the output of the above-mentioned necessary condition determination unit 39, is ON. By including condition 13 in the second correction conditions in this manner, the second correction by the second correction unit 50 is performed on the premise that the above-mentioned necessary conditions are satisfied.

Condition 14 is a condition for adjusting the frequency of execution by the second correction unit 50 based on the input second correction frequency signal Sa2. The second correction frequency signal Sa2 is a rectangular wave pulse signal that repeats an ON time for validating the second correction and an OFF time for invalidating the second correction. When the second corrected frequency signal Sa2 is input, condition 14 is determined to be satisfied during the ON time, and determined not to be satisfied during the OFF time. In condition 14, the frequency of performing the second correction can be adjusted by changing the ratio between the ON time and the OFF time of the input second correction frequency signal Sa2. The ON time and OFF time of the second corrected frequency signal Sa2 can be set by the user.

Condition 15 is a condition for determining whether the first correction permission flag for determining permission to implement the second correction is ON. For example, the control device 15 includes an operation button (not shown) that allows the user to select whether or not to perform the second correction, and when the operation button is turned on, the second correction permission flag is turned on. This allows the user to select whether or not to perform the second correction.

The second correction condition determination unit 51 determines that the second correction condition is satisfied when all of these conditions 12 to 15 are satisfied, and outputs the second correction signal S2. As described above, the second correction frequency signal Sa2, which is a rectangular wave pulse signal, is input to the second correction condition as condition 14. Therefore, the second correction signal S2 is adjusted so as to correspond to the second correction frequency signal Sa2. It is output as a rectangular wave pulse signal (flicker signal) in which ON time and OFF time are repeated.

The second correction coefficient calculation unit 52 is configured to calculate a second correction coefficient based on the second correction signal S2 output from the second correction condition determination unit 51. Here, FIG. 9 is a processing flow diagram of the second correction coefficient calculation unit 52 of FIG. 2.

In the second correction coefficient calculation unit 52, when the difference ΔV between the corrected predicted value Vp′ (predicted value Vp of the learning model M itself before the first correction) and the actual measured value Vm is input, if the difference ΔV is negative (That is, when the predicted value Vp is less than the actual value Vm), the fourth switch T4 is switched to select the upward correction value A3. The upward correction value A3 is set as a coefficient of 1 or more obtained by adding the upward correction width to the reference value "1", and is, for example, "1.01". The upward correction value A3 output from the fourth switch T4 is multiplied by a correction gain K (typically set to "1.0", but can be changed as appropriate) set in advance by the correction gain adjustment section P. After that, the second correction signal S2 outputted from the second correction condition determining section 51 described above is outputted to the fifth switch T5 at the timing when the second correction signal S2 is turned on (note that the second correction signal S2 inputted to the fifth switch T5 At the timing when the correction signal S2 is in the OFF state, the fifth switch T5 outputs the default value "1"). The output of the fifth switch T5 is multiplied by the previous value (second correction coefficient) stored in advance, and is output as the second correction coefficient Z.
Note that the second correction coefficient Z output in this way is stored in a storage unit such as a memory (not shown), and is used as the previous value in the next calculation cycle.

In this way, when the difference ΔV is negative, the second correction coefficient Z increases every calculation cycle by the upward correction width set in the upward correction value A3. Thereby, when the corrected predicted value Vp' is smaller than the actual measured value Vm, the second correction coefficient Z is calculated so that the corrected predicted value Vp' is corrected in an increasing direction and approaches the actual measured value Vm.

On the other hand, when the difference ΔV is positive (that is, when the corrected predicted value Vp' is larger than the actual measured value Vm), the fourth switch T4 is switched to select the downward correction value A4. The downward correction value A4 can be set as a coefficient less than 1 obtained by adding the downward correction width (negative value) to the reference value “1”, and is, for example, “0.99”. The downward correction value A4 output from the fourth switch T4 is multiplied by a correction gain K (typically set to "1", but can be changed as appropriate) set in advance by the correction gain adjustment section P. , is outputted to the fifth switch T5 at the timing when the second correction signal S2 outputted from the second correction condition determining section 51 described above turns ON (note that the second correction signal S2 inputted to the fifth switch T5 At the timing when S2 is in the OFF state, the fifth switch T5 outputs the default value "1"). The output of the fifth switch T5 is multiplied by the previous second correction coefficient Z stored in advance, and is output as the second correction coefficient Z.
Note that the second correction coefficient Z output in this way is stored in a storage unit such as a memory (not shown), and is used as the previous value in the next calculation cycle.

In this way, when the difference ΔV is positive, the second correction coefficient Z decreases every calculation cycle by the downward correction width set in the downward correction value A4. Thereby, when the corrected predicted value Vp' is larger than the actual measured value Vm, the second correction coefficient Z is calculated so that the corrected predicted value Vp' is corrected in a decreasing direction and approaches the actual measured value Vm.

The second correction coefficient Z calculated by the second correction coefficient calculation unit 52 is output from the sixth switch T6 and multiplied by the predicted value Vp of the learning model M when the second correction permission flag is turned ON. By doing so, a corrected predicted value Vp' that has been subjected to the second correction is obtained. Such second correction is performed by multiplying the predicted value Vp by the second correction coefficient Z, regardless of which of the first load range Lr1 to the fourth load range Lr4 the load L belongs to. Therefore, the predicted value Vp can be corrected as a whole without depending on the value of the load L, and the prediction accuracy can be effectively improved without reconstructing the learning model M.

Next, a method for creating the table Tb by the table creating section 31 will be described with reference to FIG. 10. FIG. 10 is a processing flow diagram of the table creation section 31 of FIG. 2.

To create the table Tb, first input each parameter included in the operating data received by the operating data receiving unit 30 into the learning model M, and obtain the predicted value Vp (predicted value of SO ₂ concentration at the absorption tower outlet) of the learning model M. calculate. The obtained predicted value Vp is corrected by at least one of the first correction section 40 and the second correction section 50 depending on the success or failure status of the above-mentioned necessary conditions, first correction conditions, and second correction conditions. Thereby, a corrected predicted value Vp' having higher prediction accuracy than the predicted value Vp of the learning model M is obtained. Then, a target absorbent concentration value in the absorbent liquid is calculated as an optimal value for the corrected predicted value Vp' to satisfy a preset reference value (for example, become equal to or less than the reference value). In FIG. 10, by inputting combinations of such loads L1, L2, . . . and optimal absorbent concentration values M1, M2, . It shows how a table Tb that defines relationships is created.

Then, the control target value determination unit 32 inputs the load L included in the operation data into the table Tb created by the table creation unit 31, so that the corrected predicted value Vp' can be set to a preset reference value. An absorbent concentration target value M to be satisfied is determined, and control target values for the absorbent input amount and the circulation flow rate of the absorbent corresponding to the absorbent concentration target value M are determined.

Here, the control target value determining unit 32 further corrects the absorbent concentration target value M obtained from the table Tb using a feedforward component, thereby determining the control target value regarding the amount of absorbent to be introduced. You may. FIG. 11 is a process flow diagram regarding the control target value determination process by the control target value determination unit 32 of FIG. 2. In FIG.

The control target value determination unit 32 first inputs the difference Δ between the absorbent concentration target value M and its actual measurement value Mm into the PI controller. The PI control value output from the PI controller is corrected using a feedforward correction value calculated based on operational data. Specifically, the first feedforward correction value ff1 corresponding to the exhaust gas flow rate of the combustion device 1 and the SO ₂ concentration at the absorption tower inlet, and the second feedforward value obtained by converting the input amount predicted value using conversion logic. A feedforward correction value ff is obtained by adding the correction value ff2.
The feedforward correction value ff obtained in this manner is added to the PI control value to determine the control target value regarding the amount of absorbent to be introduced. The control target value determined in this manner enables advance control using the feedforward correction value, so that the control accuracy by the control device 15 can be further improved.

In the control device 15 described above, as described above with reference to FIG. 2, the predicted value correction unit 36 corrects the predicted value Vp of the learning model M, and creates the table Tb using the corrected predicted value Vp'. The control target value determination unit 32 determines the control target value using the table Tb. Such a control device 15 can change or add the above processing by adjusting the parameters of the predicted value correction section 36 and the control target value determination section 32. For example, the predicted value correction unit 36 uses the reference value of ΔVref, the ON/OFF time of the second correction frequency signal Sa2 of the above-mentioned first correction condition Condition 10, the above-mentioned second correction condition Condition 14, and the above-mentioned upward correction value A1. And A3, the above-mentioned downward correction values A2 and A4, the correction gain K adjustable by the correction gain adjustment section P, the upper and lower limits of the first load-specific correction coefficient Y1 and the second correction coefficient Z can be changed. In addition, the control target value determining unit 32 inputs absorbent based on PI parameters (parameters related to the PI controller in FIG. 11, such as the proportionality constant K and integral time T), the output ff1 of the function Fx in FIG. 11, and the output ff2 of the conversion logic. It is possible to change or add the amount.

Here, if all the functional blocks of the control device 15 including the predicted value correction section 36 and the control target value determination section 32 are arranged in the upper control device 15A, the functional blocks are written by a program in the upper control device 15A. When changing or adding processes, at least two people are required, including a program engineer to handle the upper control device 15A and a control logic engineer to handle the lower control device 15B. Become.

On the other hand, according to the above configuration, as described above with reference to FIG. This makes it possible for only one person (an engineer who can understand the control logic) to handle the process, effectively reducing the number of personnel required to change or add processing.

Further, FIG. 12 is a processing flow diagram of the circulation pump adjustment section 33 of FIG. 2. The circulation pump adjustment section 33 adjusts the number of operating circulation pumps based on the control target value regarding the circulation flow rate of the absorption liquid determined by the control target value determination section 32. The circulation pump adjustment unit 33 adjusts the number of operating circulation pumps by issuing a command to increase the number of operating circulation pumps, for example, when the following four conditions are satisfied.
(Condition 16) When the difference Δ between the target SO2 concentration at the absorption tower outlet and the measured value Vm exceeds the allowable value and changes in the positive direction (i.e., the circulation pump is (if emergency startup is required), the difference Δ between the optimal number of circulating pumps and the current number of operating pumps is smaller than the standard value (for example, -0.5 units).
(Condition 17) After starting one circulation pump, there is no additional start-up prohibition period until it is reflected in the SO2 concentration at the outlet of the absorption tower.
(Condition 18) There must be a pump that can be started.
(Condition 19) Optimum operation AI is in use.

Further, the circulation pump adjustment unit 33 adjusts the number of operating circulation pumps by issuing a command to reduce the number of operating circulation pumps, for example, when the following four conditions are satisfied.
(Condition 16) When the difference Δ between the target value of SO2 concentration at the outlet of the absorption tower and the actual value Vm is less than the allowable value and changes in the negative direction (i.e., the difference between the target value of SO2 concentration at the outlet of the absorption tower and the actual value Vm is large and it is possible to reduce the number of pumps), the difference Δ between the optimal number of circulation pumps and the current number of operating pumps is larger than the reference value (for example, 0.5 units).
(Condition 20) Stopping after startup is not prohibited (to prevent frequent starting and stopping of the circulation pump).
(Condition 21) Stopping is not prohibited after stopping (preventing frequent starting and stopping of the circulation pump).
(Condition 22) There is a pump that can be stopped.

Next, variations in the configuration of the control device 15 described above will be described with reference to FIG. 13. 13A and 13B are block configuration diagrams showing other aspects of FIG. 2.

In the embodiment shown in FIG. 13A, the functional blocks (driving data receiving unit 30, learning model construction unit 38, table creation unit 31) included in the upper control device 15A in FIG. 2 are integrated into the lower control device 15B. That is, the lower control device 15B, except for the relearning device 15C, includes an operation data receiving section 30, a learning model construction section 38, a table creation section 31, a predicted value correction section 36, a control target value determination section 32, and a circulation pump adjustment section. 33 and an absorbent slurry supply control section 34, and all of these functional blocks are realized by control logic. Note that some of these functional blocks included in the lower control device 15B may be implemented in a program language that can be handled by the lower control device 15B.

In the embodiment shown in FIG. 13A, the functional blocks (driving data receiving section 30, learning model construction section 38, table creation section 31) that were arranged in the upper control device 15A and described by the program in the embodiment of FIG. This is realized by being written as control logic in the lower control device 15B. By consolidating various functions of the control device 15 into the lower control device 15B, for example, it becomes unnecessary to install the upper control device 15A at the site where the wet flue gas desulfurization device 10 to be controlled is located, and the control device 15 management burden can be reduced.

In the embodiment shown in FIG. 13B, an edge server 15D is provided together with the upper control device 15A and the lower control device 15B. The edge server 15D includes a data relay unit 60, and is placed between the upper control device 15A and the lower control device 15B, so that data between the upper control device 15A and the lower control device 15B, which are located geographically apart from each other, is transferred. Enables sending and receiving. In this case, the upper control device 15A controls the control state of the wet flue gas desulfurization device 10 in a remote location away from the site where the lower control device 15B is installed (the place where the wet flue gas desulfurization device 10 to be controlled is installed). It can be applied to an embodiment in which the system functions as a remote monitoring system for remotely monitoring.

Further, as shown in FIG. 13B, the wet flue gas desulfurization device 10 may constitute a remote monitoring system including, for example, an information processing device 18 that can communicate with a host control device 15A. In this remote monitoring system, remote monitoring can be performed by displaying information regarding the control state of the wet flue gas desulfurization device 10 on the display unit 70 provided in the information processing device 18.
Further, in such a system configuration, a configuration may be provided in which each processing flow in the control device 15 is executed in response to a request from the information processing device 18 via the display unit 70 of the information processing device 18.

As explained above, according to each of the above embodiments, the predicted values of the learning model used to create the table used to determine the control target value of the amount of absorbent input and the circulating flow rate of the absorbent liquid are corrected. Corrected using a coefficient. The predicted value is corrected by using the first correction coefficient calculated based on the load of the combustion device, taking into account the characteristics of the load. For example, the predicted value and the actual value are Even if the difference between the two values changes, an accurate corrected predicted value can be obtained. By creating a table using such corrected predicted values, the control target value can be determined with high accuracy, and as a result, even if there is a difference between the predicted value of the learning model and the actual measured value, Control accuracy can be obtained.

In addition, the components in the embodiments described above can be replaced with well-known components as appropriate without departing from the spirit of the present disclosure, and the embodiments described above may be combined as appropriate.

The contents described in each of the above embodiments can be understood as follows, for example.

(1) A control device for a wet flue gas desulfurization device according to one aspect includes:
A control device for a wet flue gas desulfurization device that performs desulfurization by bringing exhaust gas generated in a combustion device into gas-liquid contact with an absorption liquid in an absorption tower,
At least one parameter included in the operation data of at least one of the combustion device or the wet flue gas desulfurization device, the explanatory variable including at least the load of the combustion device, and the future sulfur dioxide concentration at the outlet of the absorption tower. a learning model construction unit for constructing a learning model using machine learning regarding the relationship with the objective variable;
Predicted value correction for correcting the predicted value of the sulfur dioxide concentration by the learning model using a correction coefficient calculated based on the difference between the predicted value and the actual measured value of the sulfur dioxide concentration at the outlet of the absorption tower. Department and
A table showing the relationship between the load of the combustion device and the absorbent concentration target value and absorbent circulation amount target value of the absorbent liquid so that the corrected predicted value corrected by the predicted value correction unit satisfies the reference value. A table creation part for creating,
The absorbent concentration target value and the absorbent circulation rate target value corresponding to the operation data are calculated based on the table, and the absorbent concentration target value and the absorbent circulation rate target value corresponding to the absorbent concentration target value and the absorbent circulation rate target value are calculated. a control target value determination unit for determining a control target value of the input amount and the circulating flow rate of the absorption liquid;
Equipped with
The predicted value correction unit includes a first correction unit that corrects the predicted value using a first correction coefficient calculated based on the load as the correction coefficient.

According to the aspect (1) above, the predicted value of the learning model used to create the table used to determine the control target value of the amount of absorbent input and the circulating flow rate of the absorbent is calculated using the correction coefficient. will be corrected. The predicted value is corrected by using the first correction coefficient calculated based on the load of the combustion device, taking into account the characteristics of the load. For example, the predicted value and the actual value are Even if the difference between the two values changes, an accurate corrected predicted value can be obtained. By creating a table using such corrected predicted values, the control target value can be determined with high accuracy, and as a result, even if there is a difference between the predicted value of the learning model and the actual measured value, Control accuracy can be obtained.

(2) In another aspect, in the aspect of (1) above,
The first correction coefficient is calculated such that the difference decreases for each load range to which the load belongs.

According to the aspect (2) above, the first correction coefficient is calculated such that the difference between the predicted value of the learning model and the actual value decreases in the load range to which the load of the combustion device belongs. Thereby, it is possible to make detailed corrections to the predicted values of the learning model according to the magnitude of the load on the combustion device, and it is possible to accurately create a table for determining the control target value.

(3) In another aspect, in the aspect (1) or (2) above,
The first correction unit calculates the first correction coefficient as the correction coefficient when a first correction condition including that a difference between the corrected predicted value and the measured value is larger than a reference value is satisfied. .

According to the aspect (3) above, the predicted value is corrected using the first correction coefficient when the first correction condition is satisfied.

(4) In another aspect, in any one of the above (1) to (3),
The correction coefficient is calculated to change at the preset rate of change so that the difference decreases every calculation cycle.

According to the aspect (4) above, the correction coefficient is calculated to change at a predetermined rate of change, so that the difference between the predicted value of the learning model and the actual value is suitably reduced. Corrections are made. Further, by setting the rate of change of the correction coefficient as a parameter that can be set in advance by the user, for example, it is possible to arbitrarily adjust the degree of correction to the predicted value.

(5) In another aspect, in any one of the above (1) to (4),
The predicted value correction section includes a second correction section that corrects the predicted value using a second correction coefficient that does not depend on the load as the correction coefficient.

According to the aspect (5) above, the predicted value of the learning model is corrected using a second correction coefficient that does not depend on the load in addition to the first correction coefficient calculated based on the load described above. By performing correction using a combination of the first correction coefficient and the second correction coefficient in this manner, it is possible to more suitably reduce the difference between the predicted value of the learning model and the actual value.

(6) In another aspect, in the aspect (5) above,
The second correction unit is configured to calculate, as the correction coefficient, when a second correction condition including that a difference between the corrected predicted value and the result of multiplying the second correction coefficient is larger than a reference value is satisfied. , calculating the second correction coefficient.

According to the aspect (6) above, the predicted value is corrected using the second correction coefficient when the second correction condition is satisfied.

(7) In another aspect, in any one of the above (1) to (6),
The control target value determination unit calculates a PI control value corresponding to the difference between the absorbent concentration target value calculated based on the table and the absorbent concentration actual value of the absorbent, based on the operation data. The control target value regarding the input amount of the absorbent is determined by correcting using the feedforward correction value.

According to the aspect (7) above, when calculating the control target value regarding the amount of absorbent input based on the absorbent concentration of the absorbent calculated based on the table, the control target value is calculated based on the operation data. By performing the correction using the feedforward correction value, it becomes possible to calculate the control target value with high accuracy.

(8) In another aspect, in any one of the above (1) to (7),
a higher-level control device whose functional blocks are composed of programs;
A lower control device that cooperates with the upper control device and whose functional blocks are composed of control logic,
At least one of the predicted value correction section and the control target value determination section is provided in the lower control device.

According to the aspect (8) above, at least one of the predicted value correction section and the control target value determination section among the functional blocks included in the control device is configured by control logic in the lower control device. When changing or adding processing by the aforementioned control device, it is necessary to make adjustments such as changing the set values in the predicted value correction section and control target value determination section. By providing it in a lower-level control device configured with logic, it becomes possible to handle it without requiring personnel who are familiar with the programs handled by the higher-level control device. Furthermore, by reducing the number of functional blocks included in the higher-level control device, it is possible to reduce the machine specifications required for the higher-level control device.

(9) In another aspect, in the aspect (8) above,
The upper control device is a main control device,
The lower-level control device is a distributed control device that is placed under the control of the higher-level control device and controls the components of the wet flue gas desulfurization device.

According to the aspect (9) above, it is suitable for a control device for a wet flue gas desulfurization device that uses a distributed control device (DCS) as a lower-level control device.

(10) In another aspect, in the aspect (8) or (9) above,
The learning model construction section, the predicted value correction section, the table creation section, and the control target value determination section are provided in the lower control device.

According to the aspect (10) above, these functional blocks may be aggregated so that they are all configured by control logic in the lower control device.

(11) In another aspect, in any one of the above (8) to (10),
A relearning device electrically connected to at least one of the upper control device or the lower control device for relearning the learning model.

According to the above aspect (11), if the accuracy is not sufficiently improved by correcting the predicted value of the learning model, it can be dealt with by relearning the learning model using the relearning device. In this case, the relearning device is configured separately from the above-mentioned higher-level control device and lower-level control device, and by electrically connecting to these control devices, the re-learning device can adjust the machine specifications required by the higher-level control device and lower-level control device. It can be suppressed.

(12) In another aspect, in any one of the above (8) to (11),
The upper control device is a remote control device for remotely monitoring the lower control device installed at the site where the wet flue gas desulfurization device is installed,
The lower-level control device can communicate with the higher-level control device via an edge server that can relay data between the higher-level control device and the lower-level control device.

According to the aspect (12) above, data relay by the edge server enables data transmission and reception between the upper control device and the lower control device. Therefore, by arranging the upper control device and the lower control device at remote locations from each other, a remote control device for the flue gas desulfurization device can be realized.

(13) A remote monitoring system according to one aspect includes:
A remote monitoring system comprising an information processing device that can communicate with a control device of a wet flue gas desulfurization device that desulfurizes exhaust gas generated in a combustion device and an absorption liquid by bringing it into gas-liquid contact in an absorption tower,
The control device includes:
In response to a request from the information processing device, at least one parameter included in the operation data of at least one of the combustion device or the wet flue gas desulfurization device, including at least the load of the combustion device, and a future a learning model construction unit that constructs a learning model by machine learning regarding the relationship between the target variable, which is the sulfur dioxide concentration at the absorption tower outlet;
a predicted value correction unit that corrects the predicted value of the sulfur dioxide concentration by the learning model using a correction coefficient calculated based on the difference between the predicted value and the actual measured value of the sulfur dioxide concentration at the outlet of the absorption tower; ,
A table showing the relationship between the load of the combustion device and the absorbent concentration target value and absorbent circulation amount target value of the absorbent liquid so that the corrected predicted value corrected by the predicted value correction unit satisfies the reference value. A table creation part to create,
The absorbent concentration target value and the absorbent circulation rate target value corresponding to the operation data are calculated based on the table, and the absorbent concentration target value and the absorbent circulation rate target value corresponding to the absorbent concentration target value and the absorbent circulation rate target value are calculated. a control target value determination unit that determines a control target value for the input amount and the circulation flow rate of the absorption liquid;
Equipped with
The predicted value correction unit includes a first correction unit that corrects the predicted value using a first correction coefficient calculated based on the load as the correction coefficient.

According to the aspect (13) above, remote monitoring is suitably performed by displaying information regarding the control state of the wet flue gas desulfurization apparatus 1 according to each of the above aspects on a display unit provided in the information processing device. be able to.

(14) A method for controlling a remote monitoring system according to one aspect includes:
A method for controlling a remote monitoring system comprising an information processing device capable of communicating with a control device of a wet flue gas desulfurization device that performs desulfurization by bringing exhaust gas generated in a combustion device into gas-liquid contact with an absorption liquid in an absorption tower, ,
The control device includes:
In response to a request from the information processing device, at least one parameter included in the operation data of at least one of the combustion device or the wet flue gas desulfurization device, including at least the load of the combustion device, and a future a learning model construction step of constructing a learning model by machine learning regarding the relationship with a target variable, which is the sulfur dioxide concentration at the absorption tower outlet;
a predicted value correction step of correcting the predicted value of the sulfur dioxide concentration by the learning model using a correction coefficient calculated based on the difference between the predicted value and the actual measured value of the sulfur dioxide concentration at the outlet of the absorption tower; ,
A table showing the relationship between the load of the combustion device and the absorbent concentration target value and absorbent circulation amount target value of the absorbent liquid so that the corrected predicted value corrected by the predicted value correction step satisfies the reference value. The table creation process to be created,
The absorbent concentration target value and the absorbent circulation rate target value corresponding to the operation data are calculated based on the table, and the absorbent concentration target value and the absorbent circulation rate target value corresponding to the absorbent concentration target value and the absorbent circulation rate target value are calculated. a control target value determining step of determining a control target value of the input amount and the circulation flow rate of the absorption liquid;
Run
The predicted value correction step includes a first correction step of correcting the predicted value using a first correction coefficient calculated based on the load as the correction coefficient.

According to the aspect (14) above, remote monitoring is suitably performed by displaying information regarding the control state of the wet flue gas desulfurization apparatus 1 according to each of the above aspects on a display unit provided in the information processing device. be able to.

(15) A method for controlling a wet flue gas desulfurization device according to one embodiment includes:
A method for controlling a wet flue gas desulfurization device that performs desulfurization by bringing exhaust gas generated in a combustion device into gas-liquid contact with an absorption liquid in an absorption tower, the method comprising:
At least one parameter included in the operation data of at least one of the combustion device or the wet flue gas desulfurization device, the explanatory variable including at least the load of the combustion device, and the future sulfur dioxide concentration at the outlet of the absorption tower. A step of constructing a learning model using machine learning regarding the relationship with the objective variable;
correcting the predicted value of the sulfur dioxide concentration by the learning model using a correction coefficient calculated based on the difference between the predicted value and the actual measured value of the sulfur dioxide concentration at the outlet of the absorption tower;
A table showing the relationship between the load of the combustion device and the absorbent concentration target value and absorbent circulation amount target value of the absorbent liquid so that the corrected predicted value corrected by the predicted value correction unit satisfies the reference value. The process of creating
The absorbent concentration target value and the absorbent circulation rate target value corresponding to the operation data are calculated based on the table, and the absorbent concentration target value and the absorbent circulation rate target value corresponding to the absorbent concentration target value and the absorbent circulation rate target value are calculated. determining control target values for the input amount and the circulating flow rate of the absorption liquid;
Equipped with
In the step of correcting the predicted value, the predicted value is corrected using a first correction coefficient calculated based on the load as the correction coefficient.

According to the aspect (15) above, the predicted value of the learning model used to create the table used to determine the control target value of the amount of absorbent input and the circulation flow rate of the absorbent liquid uses the correction coefficient. will be corrected. The predicted value is corrected by using the first correction coefficient calculated based on the load of the combustion device, taking into account the characteristics of the load. For example, the predicted value and the actual value are Even if the difference between By creating a table using such corrected predicted values, the control target value can be determined with high accuracy, and as a result, even if there is a difference between the predicted value of the learning model and the actual measured value, Control accuracy can be obtained.

1 Combustion device 2 Piping 3 Circulation piping 5 Generator 10 Wet flue gas desulfurization device 11 Absorption tower 12 Circulation pump 13 Absorbent slurry supply section 14 Gypsum recovery section 15 Control device 15A Upper control device 15B Lower control device 15C Relearning device 15D Edge server 16 Outflow piping 17 Gas analyzer 18 Information processing device 20 Operation data acquisition section 21 Absorbent slurry manufacturing equipment 22 Absorbent slurry supply piping 23 Absorbent slurry supply amount control valve 25 Gypsum separator 26 Gypsum slurry extraction piping 27 Pump 30 for extracting gypsum slurry Operation data receiving section 31 Table creating section 32 Control target value determining section 33 Circulating pump adjusting section 34 Absorbent slurry supply controlling section 36 Predicted value correcting section 38 Learning model constructing section 39 Necessary condition determining section 40 First Correction unit 41 First correction condition determination unit 42 Load-specific correction signal generation unit 42a1 to 42a4 First to fourth load range determination units 42b1 to 42b4 First to fourth load-specific correction signal output unit 43 Load-specific correction coefficient calculation unit 43a ~43d First to fourth load-based correction coefficient calculation unit 44 First correction coefficient calculation unit 50 Second correction unit 51 Second correction condition determination unit 52 Second correction coefficient calculation unit 60 Data relay unit 70 Display unit M Learning model

Claims (15)

A control device for a wet flue gas desulfurization device that performs desulfurization by bringing exhaust gas generated in a combustion device into gas-liquid contact with an absorption liquid in an absorption tower,
At least one parameter included in the operation data of at least one of the combustion device or the wet flue gas desulfurization device, the explanatory variable including at least the load of the combustion device, and the future sulfur dioxide concentration at the outlet of the absorption tower. a learning model construction unit for constructing a learning model using machine learning regarding the relationship with the objective variable;
Predicted value correction for correcting the predicted value of the sulfur dioxide concentration by the learning model using a correction coefficient calculated based on the difference between the predicted value and the actual measured value of the sulfur dioxide concentration at the outlet of the absorption tower. Department and
A table showing the relationship between the load of the combustion device and the absorbent concentration target value and absorbent circulation amount target value of the absorbent liquid so that the corrected predicted value corrected by the predicted value correction unit satisfies the reference value. A table creation part for creating,
The absorbent concentration target value and the absorbent circulation rate target value corresponding to the operation data are calculated based on the table, and the absorbent concentration target value and the absorbent circulation rate target value corresponding to the absorbent concentration target value and the absorbent circulation rate target value are calculated. a control target value determination unit for determining a control target value of the input amount and the circulation flow rate of the absorption liquid;
Equipped with
The predicted value correction unit is a control device for a wet flue gas desulfurization apparatus, including a first correction unit that corrects the predicted value using, as the correction coefficient, a first correction coefficient calculated based on the load. The control device for a wet flue gas desulfurization device according to claim 1, wherein the first correction coefficient is calculated so that the difference is reduced for each load range to which the load belongs. The first correction unit calculates the first correction coefficient as the correction coefficient when a first correction condition including that a difference between the corrected predicted value and the measured value is larger than a reference value is satisfied. A control device for a wet flue gas desulfurization device according to claim 1 or 2. The control device for a wet flue gas desulfurization device according to claim 1 or 2, wherein the correction coefficient is calculated so as to change at the preset rate of change so that the difference decreases every calculation cycle. The wet flue gas desulfurization according to claim 1 or 2, wherein the predicted value correction section includes a second correction section that corrects the predicted value using a second correction coefficient that does not depend on the load as the correction coefficient. Device control device. The second correction unit is configured to calculate, as the correction coefficient, when a second correction condition including that a difference between the corrected predicted value and the result of multiplying the second correction coefficient is larger than a reference value is satisfied. The control device for a wet flue gas desulfurization device according to claim 5, wherein the control device calculates the second correction coefficient. The control target value determining unit calculates a PI control value corresponding to the difference between the absorbent concentration target value calculated based on the table and the absorbent concentration actual value of the absorbent, based on the operation data. The control device for a wet flue gas desulfurization device according to claim 1 or 2, wherein the control target value regarding the input amount of the absorbent is determined by correcting it using a feedforward correction value. a higher-level control device whose functional blocks are composed of programs;
A lower control device that cooperates with the upper control device and whose functional blocks are composed of control logic,
The control device for a wet flue gas desulfurization device according to claim 1 or 2, wherein at least one of the predicted value correction section or the control target value determination section is provided in the lower control device. The upper control device is a main control device,
The wet flue gas desulfurization device according to claim 8, wherein the lower control device is a distributed control device that is placed under the management of the higher order control device and controls components of the wet flue gas desulfurization device. Control device. The control device for a wet flue gas desulfurization device according to claim 8 or 9, wherein the learning model construction section, the predicted value correction section, the table creation section, and the control target value determination section are provided in the lower control device. . The control device for a wet flue gas desulfurization device according to claim 8 or 9, further comprising a relearning device electrically connected to at least one of the upper control device or the lower control device and for relearning the learning model. . The upper control device is a remote control device for remotely monitoring the lower control device installed at the site where the wet flue gas desulfurization device is installed,
The wet smoke exhaust system according to claim 8 or 9, wherein the lower-level control device is capable of communicating with the higher-level control device via an edge server that can relay data between the higher-level control device and the lower-level control device. Control device for desulfurization equipment. A remote monitoring system comprising an information processing device that can communicate with a control device of a wet flue gas desulfurization device that desulfurizes exhaust gas generated in a combustion device and an absorption liquid by bringing it into gas-liquid contact in an absorption tower,
The control device includes:
In response to a request from the information processing device, at least one parameter included in the operation data of at least one of the combustion device or the wet flue gas desulfurization device, including at least the load of the combustion device, and a future a learning model construction unit that constructs a learning model by machine learning regarding the relationship between the target variable, which is the sulfur dioxide concentration at the absorption tower outlet;
a predicted value correction unit that corrects the predicted value of the sulfur dioxide concentration by the learning model using a correction coefficient calculated based on the difference between the predicted value and the actual measured value of the sulfur dioxide concentration at the outlet of the absorption tower; ,
A table showing the relationship between the load of the combustion device and the absorbent concentration target value and absorbent circulation amount target value of the absorbent liquid so that the corrected predicted value corrected by the predicted value correction unit satisfies the reference value. A table creation part to create,
The absorbent concentration target value and the absorbent circulation rate target value corresponding to the operation data are calculated based on the table, and the absorbent concentration target value and the absorbent circulation rate target value corresponding to the absorbent concentration target value and the absorbent circulation rate target value are calculated. a control target value determination unit that determines a control target value for the input amount and the circulation flow rate of the absorption liquid;
Equipped with
The predicted value correction unit includes a first correction unit that corrects the predicted value using a first correction coefficient calculated based on the load as the correction coefficient, a remote monitoring system for a wet flue gas desulfurization apparatus. . A method for controlling a remote monitoring system comprising an information processing device capable of communicating with a control device of a wet flue gas desulfurization device that performs desulfurization by bringing exhaust gas generated in a combustion device into gas-liquid contact with an absorption liquid in an absorption tower, ,
The control device includes:
In response to a request from the information processing device, at least one parameter included in the operation data of at least one of the combustion device or the wet flue gas desulfurization device, including at least the load of the combustion device, and a future a learning model construction step of constructing a learning model by machine learning regarding the relationship with a target variable, which is the sulfur dioxide concentration at the absorption tower outlet;
a predicted value correction step of correcting the predicted value of the sulfur dioxide concentration by the learning model using a correction coefficient calculated based on the difference between the predicted value and the actual measured value of the sulfur dioxide concentration at the outlet of the absorption tower; ,
A table showing the relationship between the load of the combustion device and the absorbent concentration target value and absorbent circulation amount target value of the absorbent liquid so that the corrected predicted value corrected by the predicted value correction step satisfies the reference value. The table creation process to be created,
The absorbent concentration target value and the absorbent circulation rate target value corresponding to the operation data are calculated based on the table, and the absorbent concentration target value and the absorbent circulation rate target value corresponding to the absorbent concentration target value and the absorbent circulation rate target value are calculated. a control target value determining step of determining a control target value of the input amount and the circulation flow rate of the absorption liquid;
Run
The predicted value correction step includes a first correction step of correcting the predicted value using, as the correction coefficient, a first correction coefficient calculated based on the load. A method for controlling a wet flue gas desulfurization device that performs desulfurization by bringing exhaust gas generated in a combustion device into gas-liquid contact with an absorption liquid in an absorption tower, the method comprising:
At least one parameter included in the operation data of at least one of the combustion device or the wet flue gas desulfurization device, the explanatory variable including at least the load of the combustion device, and the future sulfur dioxide concentration at the outlet of the absorption tower. A step of constructing a learning model using machine learning regarding the relationship with the objective variable;
correcting the predicted value of the sulfur dioxide concentration by the learning model using a correction coefficient calculated based on the difference between the predicted value and the actual measured value of the sulfur dioxide concentration at the outlet of the absorption tower;
A table showing the relationship between the load of the combustion device and the absorbent concentration target value and absorbent circulation amount target value of the absorbent liquid so that the corrected predicted value corrected by the predicted value correction unit satisfies the reference value. The process of creating
The absorbent concentration target value and the absorbent circulation rate target value corresponding to the operation data are calculated based on the table, and the absorbent concentration target value and the absorbent circulation rate target value corresponding to the absorbent concentration target value and the absorbent circulation rate target value are calculated. determining control target values for the input amount and the circulating flow rate of the absorption liquid;
Run
In the step of correcting the predicted value, the predicted value is corrected using a first correction coefficient calculated based on the load as the correction coefficient.

Images