Power generation plant control system, power generation plant control method, and power generation plant control program

Abstract

The present invention is related to a power generation plant control system for controlling a power generation plant capable of generating power by driving a turbine connected to a generator with steam generated by a boiler device. This system uses a prediction model constructed through machine learning using operation data of a power generation plant as learning data to predict a process value of the power generation plant corresponding to the operation data, and optimizes an operation condition of the power generation plant on the basis of the process value to determine an optimized operation condition. The optimized operation condition is stored in a storage unit in association with the operation state of the power generation plant. When it is determined that a past optimized operation condition corresponding to the operation state of the power generation plant stored in the storage unit is applicable as an operation condition, the past optimized operation condition is set as the operation condition.

Description

Power Generation Plant Control System, Power Generation Plant Control Method, and Power Generation Plant Control Program

[0001] The present disclosure relates to a power generation plant control system, a power generation plant control method, and a power generation plant control program. This application claims priority based on Japanese Patent Application No. 2024-162387 filed with the Japan Patent Office on September 19, 2024, the content of which is incorporated herein by reference.

[0002] There is known a power generation plant that generates steam by burning fossil fuels such as coal in a boiler device and drives a turbine connected to a generator with the steam to generate electricity. This type of power generation plant is operated by controlling each component device with a control system. In the control system, control is implemented by a control signal generated based on operation data acquired from the power generation plant under operating conditions set according to the operating state of the power generation plant.

[0003] The operating conditions set in the control system are, for example, to calculate an index related to plant performance from the process values of the power generation plant and optimize it so that the index becomes optimal. When the process value of the power generation plant used for such optimization of the operating conditions is a parameter that cannot be measured actually, it may be obtained as a calculated predicted value using a prediction model that simulates the behavior of the power generation plant. For example, in Patent Document 1, in a system that targets a coal-fired power plant, it is proposed to improve the prediction accuracy of the process value under each operating condition by making the operation data used for constructing the prediction model switchable for each operating condition.

[0004] Patent No. 7222943

[0005] Patent Document 1, mentioned above, allows switching of the operating data used to construct the prediction model for each operating condition. When this technology is applied to the control system of the aforementioned power plant, each time the operating state of the power plant changes, the prediction model is reconstructed using the operating data, and the operating conditions are optimized using the process values ​​predicted by the prediction model. Such reconstruction and optimization processes of the prediction model are computationally intensive and require a considerable amount of time to complete. Therefore, when the operating state of the power plant changes, while the system is calculating and optimizing the operating conditions to correspond to the changed operating state, the operating conditions optimized for the previous operating state continue to be applied. During this time, the power plant is controlled under operating conditions that are not optimized for the operating state, which may lead to a decrease in plant performance.

[0006] Furthermore, in recent years, renewable energy power generation facilities capable of generating electricity using renewable energy have become increasingly widespread. Since the amount of electricity generated by such renewable energy power generation facilities is greatly affected by environmental conditions, other power plants, such as thermal power plants connected to the same power grid, are required to function as buffers to absorb the changes in the amount of electricity generated by renewable energy power generation facilities. In this case, other power plants are required to respond flexibly to power generation commands from external sources, such as a central control room that manages the common power grid, by performing partial load operation in addition to conventional rated load operation. In this case, as described above, if optimization processing of operating conditions is performed every time the operating state of the power plant changes, there is a risk that the plant performance will deteriorate because the power plant will be controlled under operating conditions that are not optimized for the operating state until the optimization processing is completed.

[0007] At least one embodiment of this disclosure has been made in view of the above circumstances and aims to provide a power plant control system, a power plant control method, and a power plant control program that can improve plant performance by controlling the power plant based on operating conditions suitable for the operating state.

[0008] A power plant control system according to at least one embodiment of the present disclosure is a power plant control system for controlling a power plant capable of generating electricity by driving a turbine connected to a generator with steam generated by a boiler device, in order to solve the above problems, comprising: an operation data acquisition unit for acquiring operation data of the power plant; an optimization unit for determining optimized operation conditions by predicting process values ​​of the power plant corresponding to the operation data using a prediction model constructed by machine learning with the operation data as training data, and optimizing the operation conditions of the power plant based on the process values; a storage unit for storing the optimized operation conditions in association with the operating state of the power plant; and an operation condition setting unit for setting the operation conditions of the power plant, wherein the operation condition setting unit sets the past optimized operation conditions as operation conditions when it is determined that past optimized operation conditions corresponding to the operating state of the power plant stored in the storage unit are applicable as operation conditions.

[0009] A power plant control method according to at least one embodiment of the present disclosure is a power plant control method for controlling a power plant capable of generating electricity by driving a turbine connected to a generator with steam generated by a boiler device, in order to solve the above problems, comprising: a step of acquiring operating data of the power plant; a step of predicting process values ​​of the power plant corresponding to the operating data using a predictive model constructed by machine learning with the operating data as training data, and determining optimized operating conditions by optimizing the operating conditions of the power plant based on the process values; a step of storing the optimized operating conditions in a storage unit in association with the operating state of the power plant; and a step of setting the operating conditions of the power plant, wherein in the step of setting the operating conditions, if it is determined that past optimized operating conditions corresponding to the operating state of the power plant stored in the storage unit are applicable as the operating conditions, the past optimized operating conditions are set as the operating conditions.

[0010] A power plant control program according to at least one embodiment of the present disclosure is a power plant control program for controlling a power plant capable of generating electricity by driving a turbine connected to a generator with steam generated by a boiler device, in order to solve the above problems, wherein a computer device is capable of performing the following steps: acquiring operating data of the power plant; predicting process values ​​of the power plant corresponding to the operating data using a prediction model constructed by machine learning with the operating data as training data, and determining optimized operating conditions by optimizing the operating conditions of the power plant based on the process values; storing the optimized operating conditions in a storage unit in association with the operating state of the power plant; and setting the operating conditions of the power plant, wherein in the step of setting the operating conditions, if it is determined that past optimized operating conditions corresponding to the operating state of the power plant stored in the storage unit are applicable as the operating conditions, the past optimized operating conditions are set as the operating conditions.

[0011] According to at least one embodiment of this disclosure, a power plant control system, a power plant control method, and a power plant control program can be provided that can improve plant performance by controlling the power plant based on operating conditions suitable for the operating state.

[0012] This is a schematic diagram of a power plant control system according to one embodiment. This is a flowchart of a power plant control method according to reference technology. This is an example of a pattern of change in the operating state of a power plant. This is a flowchart of a control method according to one embodiment. This is a flowchart of a control method according to one embodiment together with Figure 4A. This is a modified example of Figure 4B. This is another example of a pattern of change in the operating state of a power plant. This is a modified example of Figure 4B. This is an explanatory diagram of a method for predicting past optimization operation conditions corresponding to the operating state. This is an explanatory diagram of a method for predicting past optimization operation conditions corresponding to the operating state. This is a modified example of Figure 4A. This is an example of an operating schedule specified in step S222 of Figure 8.

[0013] Hereinafter, several embodiments of this disclosure will be described with reference to the attached drawings. However, the dimensions, materials, shapes, relative arrangements, etc., of the configurations described as embodiments or shown in the drawings are not intended to limit the scope of this disclosure, but are merely illustrative examples.

[0014] Figure 1 is a schematic diagram of a power plant control system 1 according to one embodiment. The power plant control system 1 comprises a power plant 2 to be controlled, a control device 100 for controlling the power plant 2, and a control support device 200 for assisting the control of the control device 100.

[0015] The power plant 2 comprises a boiler 4, a steam turbine 6, and a generator 8. The boiler 4 is configured to generate steam S by heating feedwater supplied from the feedwater system using combustion gas produced by burning fuel F supplied from the fuel supply system 10 and air A supplied from the air supply system 12. The steam S, superheated to a high temperature, is sent to the steam turbine 6 as a working fluid through a water wall and superheater, which constitute the boiler (not shown).

[0016] A generator 8 is connected to the output shaft of the steam turbine 6. The generator 8 generates electricity by being driven by the power output from the steam turbine 6. The electricity generated by the generator 8 is supplied to a power system (not shown). The steam that has finished working in the steam turbine 6 is returned to condensate by the condenser 16. The condensate produced in the condenser 16 is sent to the boiler 4 via the feedwater pump 18 and used for steam generation.

[0017] Furthermore, the power plant 2 is equipped with sensors to monitor the status of each component of the power plant 2. Specifically, a fuel temperature sensor 20 is located at the outlet of the fuel supply system 10 to detect the temperature Tf of the fuel F supplied by the fuel supply system 10. An air temperature sensor 22 is located at the outlet of the air supply system 12 to detect the temperature Ta of the air A supplied by the air supply system 12. A steam temperature sensor 24 and a steam pressure sensor 26 are installed at the inlet of the steam turbine 6 to detect the temperature Ts and pressure Ps of the steam S (steam superheated in the superheater) introduced into the steam turbine 6, respectively. A condensate temperature sensor 28 is installed at the outlet of the condenser 16 to detect the temperature Tk of the condensate produced in the condenser 16.

[0018] The control device 100 is a control unit for controlling the power plant 2 having the above configuration. The control device 100 acquires parameters related to the operating state of the power plant 2 (hereinafter referred to as "operating parameters" as appropriate) and controls the power plant 2 by transmitting control signals generated by calculation processing using the acquired operating parameters to the power plant 2. The operating parameters may include calculation intermediate values ​​and control signals generated by the control device 100 based on the results detected by the aforementioned sensors installed in the power plant 2, as well as power generation command values ​​D that can be received from an external source (e.g., a central control room).

[0019] The control device 100 generates control signals corresponding to the operating conditions of the power plant 2 based on operating conditions that can be set by the control support device 200 described later. These operating conditions are set by the control support device 200 for each operating state of the power plant 2, so that the control of the power plant 2 by the control device 100 can be optimized according to the operating state of the power plant 2. In this embodiment, as an example of the control performed by the control device 100, we will describe the case in which a control signal is generated to control the feedwater pump 18 based on operating parameters so that the amount of power generated in the generator 8 becomes the power generation command value D input to the control device 100. The detailed configuration of the control device 100 will follow known examples and will be omitted here.

[0020] The control support device 200 is configured to support the control of the control device 100, and in particular has the function of setting operating conditions in the control device 100. Specifically, the control support device 200 includes an operation data acquisition unit 202, a prediction model construction unit 204, an optimization unit 206, a storage unit 208, and an operating condition setting unit 210.

[0021] The operation data acquisition unit 202 is configured to acquire operation data of the power plant 2. As described above, the operation data includes calculation intermediate values ​​and control signals generated by the control device 100 based on the results detected by the aforementioned sensors installed in the power plant 2, as well as power generation command values ​​D that can be received from an external source (e.g., a central control room). In this embodiment, the operation data acquisition unit 202 can acquire operation data via the control device 100, but it may also acquire operation data directly from the power plant 2.

[0022] The prediction model construction unit 204 is configured to construct a prediction model for predicting process values ​​of the power plant 2 corresponding to the operating data, using machine learning with the operating data as training data. This machine learning is carried out according to a machine learning algorithm. While linear regression, neural networks, or random forests are possible, they are not limited to these machine learning algorithms.

[0023] The optimization unit 206 is configured to use the prediction model constructed by the prediction model construction unit 204 to predict process values ​​of the power plant 2 corresponding to the operating data, and to optimize the operating conditions of the power plant 2 based on those process values. The optimization of the operating conditions is performed, for example, by calculating an index corresponding to the performance of the power plant 2 based on the process values ​​predicted using the prediction model M, and optimizing the index to be the best possible.

[0024] Furthermore, the index used in the optimization process can be defined from various perspectives. For example, if the performance of power plant 2 is considered to be the nitrogen oxide concentration contained in the exhaust gas of boiler 4, it may be an estimated nitrogen oxide concentration calculated from process values ​​predicted using a prediction model. Alternatively, if the performance of power plant 2 is considered to be the operating cost, it may be a cost evaluation index calculated from process values ​​predicted using a prediction model.

[0025] The memory unit 208 is configured to store the operation data acquired by the operation data acquisition unit 202 and the operation conditions optimized by the optimization unit 206 (hereinafter referred to as "optimized operation conditions") in association with the operating state of the power plant 2. In this embodiment, the operating state of the power plant 2 is defined by a power generation command value D, which is a type of operation parameter, and each optimized operation condition is stored in the memory unit 208 in association with the corresponding power generation command value D. The memory unit 208 stores the optimized operation conditions for each operating state, and the corresponding optimized operation conditions can be retrieved based on the operating state by external access. In the following description, the optimized operation conditions stored in the memory unit 208 will be referred to as "past optimized operation conditions" as appropriate.

[0026] The operation condition setting unit 210 is configured to set operation conditions for the control device 100. The operation condition setting unit 210 may set the optimized operation conditions calculated by the optimization unit 206 as the operation conditions, or it may set past optimized operation conditions stored in the storage unit 208. In the latter case, the operation condition setting unit 210 searches the storage unit 208 for past optimized operation conditions corresponding to the operating state of the power plant 2, and sets the past optimized operation conditions obtained as a search result as the operation conditions for the control device 100.

[0027] Next, a power plant control method implemented using the power plant control system 1 having the above configuration will be described. Here, we will first describe a power plant control method relating to reference technology with reference to Figure 2. Figure 2 is a flowchart of a power plant control method relating to reference technology.

[0028] First, the operation data acquisition unit 202 acquires operation data of the power plant 2 from the control device 100 (step S100). The operation data acquired in step S100 includes, as described above, the operation parameters that can be obtained from the various sensors installed in the power plant 2, the calculation intermediate values ​​and control signals generated by the control device 100, and the power generation command value D that can be received by the control device 100 from an external source (for example, a central control room).

[0029] Next, preprocessing is performed on the operating data acquired in step S100 (step S101). This preprocessing is a calculation process that uses the operating data to calculate parameters for determining the success or failure of the operating data setting conditions in the subsequent step S102. Examples of such parameters include the amount of change related to parameters included in the operating data over a predetermined period in step S100 (for example, the output of the generator 8 or the temperature of the steam S).

[0030] Next, the success or failure of the operation data setting conditions is determined using the parameters obtained in the preprocessing step S101 (step S102). The operation data setting conditions are conditions for determining whether the operation data acquired from the power plant 2 is in a set state, and are determined based on the parameters obtained in the preprocessing step S102. For example, if the parameters obtained in the preprocessing step S101 are fluctuations related to the output of the generator 8 and the temperature of the steam S over a predetermined period, then if these fluctuations are below a preset reference value, the operation data is considered to be in a set state, and the operation data setting conditions are determined to be met. If the operation data setting conditions are met (step S102: YES), the operation data acquired by the operation data acquisition unit 102 is stored in the storage unit 208 (step S103). If the operation data setting conditions are not met (step S102: NO), the process returns to step S100, and the acquisition of operation data is restarted.

[0031] Next, the success or failure of the prediction model construction conditions is determined (step S104). The prediction model construction conditions are the conditions for determining whether or not it is necessary to construct a new prediction model, and the success or failure of these conditions can be determined based on several criteria.

[0032] Here, we will exemplify some criteria for determining whether the prediction model construction conditions are met. First, if the memory unit 208 does not store a prediction model corresponding to the operating state of the power plant 2, it is determined that the prediction model construction conditions are met because a new prediction model needs to be constructed. Also, even if the memory unit 208 stores a prediction model corresponding to the operating state of the power plant 2, if a sufficiently long period of time has passed since the construction of that prediction model, it is determined that the prediction model construction conditions are met because a new prediction model needs to be constructed to ensure the reliability of the prediction model. Furthermore, even if the memory unit 208 stores a prediction model corresponding to the operating state of the power plant 2, if the prediction accuracy of that prediction model falls below a predetermined value, it is determined that the prediction model construction conditions are met because the prediction model needs to be reconstructed.

[0033] If the conditions for constructing a prediction model are met (step S104), the prediction model construction unit 204 constructs a prediction model using the operating data stored in the storage unit 108 in step S103 (step S105). In step S105, machine learning is performed using the operating data stored in step S103 as training data to construct a prediction model corresponding to the operating state of the power plant 2. The prediction model constructed in step S105 is stored in the storage unit 108 in a readable format in association with the operating state of the power plant 2.

[0034] Next, the optimization unit 206 reads a prediction model corresponding to the operating state of the power plant 2 from the storage unit 108 and performs an optimization process using the prediction model to determine the optimization operation conditions (step S106). In the optimization process of step S106, the process values ​​of the power plant 2 are predicted by inputting operating data obtained from the power plant 2 to the prediction model, and the optimization operation conditions are determined so that the index corresponding to the performance of the power plant 2, calculated using the process values, is the best possible.

[0035] Next, the operation condition setting unit 210 sets the optimized operation conditions calculated in step S106 as operation conditions in the control device 100 (step S107). As a result, the control device 100 generates control signals based on the operation data acquired from the power plant 2 according to the optimized operation conditions corresponding to the operating state, thereby enabling optimized plant control according to the operating state.

[0036] Here, Figure 3 shows an example of a pattern of changes in the operating state of power plant 2. In this example, the operating state of power plant 2 is initially in operating state A, which corresponds to the power generation command value D1. It begins to change at time t1 and transitions to operating state B, which corresponds to the power generation command value D2, at time t2. Subsequently, it begins to change again at time t3 and transitions back to the original operating state A (hereinafter referred to as "operating state A'" when distinguishing it from operating state A before time t1) at time t4.

[0037] Furthermore, the periods between times t1 and t2, and between times t3 and t4, are transient periods during which the operating state changes. In the example shown in Figure 3, the rate of change of the operating state during the transient period is set to a constant value, but this is not particularly limited.

[0038] In the reference technology shown in Figure 2, whenever the operating state changes, the optimization operating conditions are calculated in step S106 using an optimization calculation with a prediction model. However, since the calculation of the optimization operating conditions is a computationally intensive process, it takes time for the optimization operating conditions to be calculated. Therefore, when the operating state changes, the control device 100 has the operating conditions set before the operating state changed, which may lead to a decrease in plant performance during that period. This problem can be suitably resolved by the control method described below.

[0039] Figures 4A and 4B are flowcharts showing a control method according to one embodiment. First, steps S200 to S203 shown in Figure 4A of the control method in this embodiment are the same as steps S100 to S103 in the reference technology shown in Figure 2.

[0040] Next, the operating status of the power plant 2 is identified by analyzing the operating data accumulated in step S203 (step S204). In step S204, the operating status of the power plant 2 is identified by analyzing each operating parameter (for example, power generation command value D, etc.) included in the operating data.

[0041] Next, similar to step S104 described above, the success or failure of the prediction model construction conditions is determined (step S205). If the prediction model construction conditions are met (step S205: YES), a new prediction model is constructed, similar to step S105 described above (step S206). On the other hand, if the prediction model construction conditions are not met (step S205: NO), the prediction model is not constructed.

[0042] Next, moving on to FIG. 4B, it is determined whether the past optimization operation conditions stored in the storage unit 208 can be applied (step S207). The determination in step S207 can be made based on several criteria. For example, when the past optimization operation conditions cannot be applied, such as when the storage unit 208 does not contain the past optimization operation conditions (step S207: NO), the process proceeds to step S211. While the optimization operation conditions are being calculated in step S211, the base operation conditions are applied.

[0043] Here, several exemplary determination criteria in step S207 will be described. First, by accessing the storage unit 208, if the past optimization operation conditions corresponding to the operating state specified in step S204 are not stored in the storage unit 208, it is determined that the past optimization operation conditions cannot be applied.

[0044] Also, even when the past optimization operation conditions corresponding to the operating state specified in step S204 are stored in the storage unit 208, if a sufficiently long period has elapsed since the calculation of the past optimization operation conditions, it is determined that the past optimization operation conditions are old and cannot be applied to the setting of the operation conditions. This can preferably prevent the plant performance from deteriorating due to the setting of old past optimization operation conditions with poor reliability as the operation conditions.

[0045] Also, even when the optimization operation conditions corresponding to the operating state specified in step S204 are stored in the storage unit 208, if the operating conditions of the power generation plant 2 (for example, the properties of the fuel (calorific value, moisture, ash, various components contained in other fuels (carbon, hydrogen, oxygen, etc.))) have changed compared to the time when the past optimization operation conditions were calculated, it is determined that the past optimization operation conditions cannot be applied to the setting of the operation conditions. This can preferably prevent the plant performance from deteriorating due to the past optimization operation conditions calculated under different operating conditions.

[0046] When it is determined that the past optimization operation conditions can be applied as setting conditions based on such criteria (step S207: YES), the operation condition setting unit 210 reads the past optimization operation conditions from the storage unit 208 (step S208), and the operation condition setting unit 210 sets the past optimization operation conditions read in step S208 as operation conditions in the control device 100 (step S209).

[0047] Subsequently, it is determined whether the operating state of the power generation plant 2 controlled by the control device 100 to which the past optimization operation conditions are set in step S209 is normal (step S210). The determination in step S210 can be made from several viewpoints. For example, if an alarm is generated due to the operating parameters of the power generation plant 2 controlled by the control device 100 to which the past optimization operation conditions are set deviating from the allowable range, it is determined that the operating state of the power generation plant 2 is not normal. Also, if an evaluation index (such as an index related to cost or performance) calculated based on the operating parameters of the power generation plant 2 controlled by the control device 100 to which the past optimization operation conditions are set deviates from the allowable range, it is determined that the operating state is not normal.

[0048] As a result, when the operating state of the power generation plant 2 is normal (step S210: YES), the calculation of the optimization operation conditions corresponding to the operating state is performed while the past optimization operation conditions are set in the control device 100 (step S211). Since the calculation of the optimization operation conditions in this step is a process with a relatively large computational burden, it takes a certain amount of time until the completion of the calculation of the optimization operation conditions in step S211. At this time, if the previous operation conditions are still set in the control device 100, there is a risk that the plant performance will deteriorate as in the above-described prior art. In contrast, in this embodiment, since the past optimization operation conditions corresponding to the operating state of the power generation plant 2 are set in step S209, it is possible to preferably suppress the deterioration of the plant performance as in the above-described prior art.

[0049] Once the calculation of the optimization operation conditions is completed in step S211, the operation condition setting unit 210 sets the optimization operation conditions calculated in step S211 to the control device 100 (step S212). As a result, the previously set optimization operation conditions in the control device 100 are overwritten with the newly calculated optimization operation conditions.

[0050] Then, similar to step S210 described above, it is determined again whether the operating state of the power plant 2 controlled by the control device 100 with the optimized operating conditions set is normal (step S213). If it is determined that the operating state of the power plant 2 is normal (step S213: YES), the operating state of the power plant 2 becomes optimized by the control device 100 with the optimized operating conditions set.

[0051] On the other hand, if the operating state of the power plant 2 is not normal (step S210: NO), the operation condition setting unit 210 sets the control device 100 with base operating conditions, which are pre-prepared as standard operating conditions, instead of the past optimized operating conditions set in step S209 (step S215). In this way, if the past optimized operating conditions do not allow for normal operation control of the power plant 2, the control state of the power plant 2 can be maintained normally by switching to operation control based on the base operating conditions (even in this case, the optimized operating conditions are calculated in the subsequent step S211, but until the calculation of the optimized operating conditions is completed, the plant performance can be improved to some extent by controlling based on the base operating conditions).

[0052] Furthermore, if it is determined that the operating state of the power plant 2 based on the optimized operating conditions set in step S212 is not normal (step S213: NO), then, as in step S215 described above, base operating conditions are set in the control device 100 (step S216). This means that if the power plant 2 cannot be controlled normally under the optimized operating conditions set in step S212, the control state of the power plant 2 can be maintained normally by switching to operation control based on the base operating conditions.

[0053] In the operating state change pattern illustrated in Figure 3, operating condition A' is the same as the past operating state A. Therefore, at time t4 when the operating state of the power plant 2 becomes operating condition A', the memory unit 208 contains past optimization operation conditions corresponding to operating state A. Accordingly, in this case, while the optimization process using the prediction model calculates new optimization operation conditions, the past optimization operation conditions obtained from the memory unit 208 are temporarily set in the control device 100, thereby effectively suppressing the deterioration of plant performance during that period.

[0054] Figure 4C is a modified version of Figure 4B. In this modified version, once the calculation of the optimized operating conditions is completed in step S211, in step S212', the optimized operating conditions calculated in step S211 are set in the control device 100 by the operating condition setting unit 210, and the set values ​​are stored in the storage unit 208. Since it may take time from setting the optimized operating conditions calculated in step S211 in the control device 100 until it is determined whether the operating state is normal or not (for example, it may take time from setting new optimized operating conditions until the operating state stabilizes), it is not easy to determine in real time and with accuracy whether the optimized operating conditions are appropriate.

[0055] In this modified example, the optimization operation conditions calculated in step S211 are temporarily stored in the storage unit 208 in step S212, allowing for subsequent scrutiny of whether the optimization operation conditions stored in the storage unit 208 are appropriate. For example, if the scrutiny reveals that the optimization operation conditions are inappropriate, they can be deleted as appropriate.

[0056] In the modified example shown in Figure 4C, the optimization operation conditions are stored in the storage unit 208 in step S212', so step S214 in Figure 4B is unnecessary.

[0057] Figure 5 shows another example of a change pattern in the operating state of the power plant 2. In this example, the change pattern is the same as shown in Figure 3 until time t3, but at time t4, it transitions to operating state C, which corresponds to the power generation command value D3. Operating state C is different from operating state A, which corresponds to the power generation command value D1, and operating state B, which corresponds to the power generation command value D2. At time t4, the storage unit 208 stores the optimization operation conditions corresponding to operating states A and B as optimization operation conditions corresponding to the previous operating states of the power plant 2, but it does not store the optimization operation conditions corresponding to the new operating state C. Therefore, in the above embodiment, in step S207, it is determined that the past optimization operation conditions cannot be applied as the optimization operation conditions corresponding to operating state C.

[0058] Here, Figure 6 is a modified version of Figure 4B. In this modified control method, the optimization unit 206 may calculate past optimization operation conditions corresponding to the operating state of the power plant 2 based on at least one past optimization operation condition corresponding to other operating states. In this case, even if the storage unit 208 does not store past optimization operation conditions corresponding to operating state C, it is possible to predict past optimization operation conditions corresponding to operating state C based on past optimization operation conditions corresponding to at least one of operating states A or B.

[0059] In this modified example, compared to Figure 4B, if it is determined that past optimization operating conditions cannot be applied (step S207: NO), it is further determined whether past optimization operating conditions corresponding to the operating state of the power plant 2 can be predicted based on past optimization operating conditions corresponding to other operating states (step S220). In the example shown in Figure 5, when the power plant 2 is in operating state C, the memory unit 208 stores at least past optimization operating conditions corresponding to the previous operating states A or B. In step S220, it is determined whether past optimization operating conditions corresponding to operating state C can be predicted based on past optimization operating conditions corresponding to at least one of the other operating states A or B.

[0060] The prediction of past optimization operating conditions corresponding to operating state C can be performed, for example, by the following method. Figures 7A and 7B are explanatory diagrams regarding the method for predicting past optimization operating conditions corresponding to operating state C.

[0061] In Figure 7A, past optimization operating conditions corresponding to other operating conditions A and B are known, and by linear interpolation of these, past optimization operating conditions corresponding to the unknown operating condition C can be predicted. In Figure 7B, a function FX is shown that defines the relationship between the generator output, boiler load, etc., and the parameter P included in the operating conditions in the base operating conditions which are prepared in advance as standard operating conditions for power plant 2. By applying this function FX to the parameter Pb corresponding to the generator output and boiler load in other operating conditions B (for example, by translating the function FX so that it passes through operating condition B), the parameter Pc corresponding to the generator output and boiler load in operating condition C can be predicted, and the optimization operating conditions corresponding to operating condition C can be predicted.

[0062] Furthermore, as described above, it is preferable that the past optimization operating conditions under other operating conditions used to predict the past optimization operating conditions under the operating conditions of power plant 2 are also data from within a predetermined time period from the time of calculation.

[0063] In step S220, it is determined whether or not it is possible to predict past optimization operation conditions using this method. If it is determined that it is possible to predict past optimization operation conditions (step S220: YES), the prediction of past optimization operation conditions is performed using the above method (step S221), and set as operation conditions in the control device 100 (step S209). Thus, in this embodiment, even if it is determined in step S207 that past optimization operation conditions cannot be applied, if it is possible to predict those past optimization operation conditions, the predicted past optimization operation conditions can be set as operation conditions.

[0064] On the other hand, if past optimization operation conditions cannot be predicted (step S220: NO), the process proceeds to step S211, resulting in the same configuration as the embodiment described above with reference to Figures 4A and 4B.

[0065] As described above, the optimization unit 206 uses the prediction model constructed in the prediction model construction unit 204 to predict the process values ​​of the power plant 2, and performs optimization calculations so that the index calculated using these process values ​​is the best possible. In this optimization calculation, the optimization operation conditions are determined by searching within a range in which a predetermined likelihood can be secured for the process values ​​relative to the acceptable range.

[0066] In some embodiments, the optimization unit 206 may determine, for each operating state of the power plant 2, a first optimization operating condition having a relatively small first likelihood relative to the allowable range, and a second optimization operating condition having a second likelihood greater than the first likelihood. The first optimization operating condition is an optimization operating condition that prioritizes economic efficiency under stable operating conditions because it has a small likelihood relative to the allowable range, while the second optimization operating condition is an optimization operating condition that prioritizes controllability under transient conditions because it has a large likelihood relative to the allowable range.

[0067] The first and second optimization operating conditions are stored in the memory unit 208 in association with the corresponding operating state. In other words, the memory unit 208 stores two types of optimization operating conditions (first and second optimization operating conditions) for each operating state. The operation condition setting unit 210 can then read either the first or second optimization operating conditions from the memory unit 208 as past optimization operating conditions corresponding to the operating state, and set them as operating conditions in the control device 100.

[0068] Here, we will specifically describe a control method that uses either the first or second optimization operating condition as the optimization operating condition. Figure 8 is a modified version of Figure 4A (the latter half of the control method related to this modified version is the same as that of Figure 4B, so the explanation is omitted).

[0069] In this modified example, after performing steps up to S204, as in the previously described embodiment, the operating schedule for the power plant 2 is determined (step S222). In the operating schedule, for example, the time-series changes of the power generation command value D input to the control device 100 from an external source are defined as the operating schedule, and in step S222, the future pattern of changes in the operating state is grasped based on the operating schedule.

[0070] Next, based on the operating schedule identified in step S222, it is determined whether the future operating state of the power plant 2 is in a stable state (step S223). In step S223, for example, it is possible to determine whether the state is stable based on whether the amount of fluctuation in the future operating state exceeds a preset threshold, based on the operating schedule identified in step S222.

[0071] If it is determined that the future operating state is a stable state (step S223: YES), the process proceeds to step S205, as in the embodiment described above. In this case, the past optimization operating conditions read out in step S208 (see Figure 4B) are the first optimization operating conditions which have low likelihood and excellent economic performance. As a result, when the operating state is a stable state, operation prioritizing economic efficiency becomes possible.

[0072] On the other hand, if it is determined that the system is not in a stable state (step S223: NO), it is determined whether a second optimization operation condition with high likelihood and excellent controllability can be applied as a past optimization operation condition (step S224). In step S224, the determination can be made based on several criteria. For example, if a second optimization operation condition corresponding to the operating state is stored in the memory unit 208, it is determined that the second optimization operation condition can be applied. Even if a second optimization operation condition corresponding to the operating state is not stored in the memory unit 208, if it is possible to predict the second optimization operation condition corresponding to the transient state based on the optimization operation conditions corresponding to the stable states before and after the transient state, it is determined that the second optimization operation condition can be applied (for example, the second optimization operation condition corresponding to the transient state can be predicted by apportioning the second optimization operation conditions corresponding to the stable states before and after the transient state).

[0073] As a result, if it is determined that the second optimization operating condition is applicable (step S224: YES), the process returns to step S208. This ensures the stability of the control state of the power plant 2 by ensuring a high likelihood when the operating state is in a transient state, as the operating state is not stable.

[0074] Now, referring to Figure 9, we will explain using power plant 2, which is operated according to a specific operating schedule, as an example. Figure 9 is an example of an operating schedule specified in step S222 of Figure 8.

[0075] In this example, the operating state is constant before time t1, but changes from time t1 to time t2, and then settles back into a constant state at time t2. In step S223, based on the behavior of the operating state in the operating schedule identified in step S222, a stable state in which the operating state is constant and a transient state in which the operating state changes are identified.

[0076] In the embodiment shown in Figure 9, in particular, when identifying transient states from the operating schedule, a predetermined margin is taken into consideration. In the operating schedule, the operating state actually changes from time t1 to time t2, but when identifying the operating state, the transient state is identified as the period from time ta, which is a predetermined margin T1 before time t1, to time tb, which is a predetermined margin T2 after time t2. By identifying the transient state broadly by considering this margin, controllability is prioritized during periods when the operating state is prone to instability, including immediately before and after the transient state, while economic efficiency is prioritized during stable settling states. As a result, overall, it is possible to improve the plant performance while considering the balance between the controllability and economic efficiency of the power plant.

[0077] For example, in power plant 2, during transient states where the load on boiler 4 changes, an imbalance between fuel F from fuel supply system 10 and feedwater from feedwater system can cause parameters such as metal temperature to deviate from the acceptable range, potentially triggering various alarms. In recent years, the use of renewable energy power generation facilities has become widespread, but the amount of electricity generated from renewable energy is easily affected by weather conditions. Therefore, power plants 2 connected to a common power grid are increasingly required to have a buffer function to absorb these fluctuations. In this case, power plant 2 may operate by increasing the load change rate of boiler 4 or lowering the minimum load value. When such operation is performed using power plant 2, which is primarily designed for steady-state operation, the possibility of control parameters deviating from the acceptable range increases during transient states. In this embodiment, as described above, by selecting a second optimization operating condition with a high likelihood during transient states where the possibility of control parameters deviating from the acceptable range increases, the risk of such deviations from the acceptable range can be effectively suppressed. On the other hand, in stable state settings, by selecting a first optimization operating condition with a low likelihood, the economic efficiency of operation can be improved.

[0078] In each of the above embodiments, the storage unit 208 stores the optimization operation conditions for each operating state, and the operation condition setting unit 210 reads the optimization operation conditions corresponding to the operating state of the power plant 2 from the storage unit 208 and sets them in the control device 100. The optimization operation conditions may be prepared in the storage unit 208 to correspond to each operating state, but they may also be prepared for each range of parameters that define the operating state.

[0079] For example, if the boiler load is used as a parameter to define the operating state, the memory unit 208 may store optimization operating conditions for each load range with a certain width. For example, if the memory unit 208 stores optimization operating conditions for each numerical value indicating the load, such as 100%, 99%, 98%, etc., the number of holding circuits in the memory unit 208 becomes enormous, and even if the change in the operating state is gradual, the setting of the operating conditions may continue to change during operation, potentially impairing the stability of the operation. Therefore, in this embodiment, by defining the operating state for each load range with a certain width (for example, 30-40%, 40-65%, 65-85%, 85-100%, etc.), the size of the holding circuits in the memory unit 208 can be reduced while ensuring the stability of the operation. Such load range settings can be efficiently performed, for example, by dividing the load into frequently used load ranges based on the operational history of the power plant 2.

[0080] Furthermore, the parameters defining the operating state can broadly include parameters that affect the operating state of the power plant 2. For example, the operating state may be defined by the power generation command value D and the load as described above, but in addition to these, parameters related to the type of fuel handled in the fuel supply system 10 and the operating state of the fuel supply system 10 may also be included.

[0081] For example, the fuel supply system 10 handles coal fuel, biomass fuel, or pulverized fuel produced by crushing a mixture of coal and biomass fuel as fuel F supplied to the boiler 4. In addition, ammonia may be used as a liquid (or gaseous) fuel and may be supplied to the boiler 4 together with coal fuel or biomass fuel. The following example describes the case where only coal is used. The fuel supply system 10 includes a coal fuel source for supplying coal fuel and multiple mills for producing pulverized coal fuel by crushing the coal fuel from the coal fuel source. In this case, parameters that define the operating state may include the type of coal fuel supplied from the coal fuel source and the number of operating mills for crushing the coal fuel. In this case, depending on the coal type, the number of operating mills may differ even at the same boiler load depending on the calorie level. Also, when raising or lowering the boiler load, each mill is turned on / off, so the number of operating mills may not be the same even at the same boiler load. Therefore, by defining the operating state based on the type of coal and the number of operating mills, the scale of the storage circuit in the memory unit 208 can be reduced while ensuring operational stability.

[0082] Furthermore, it is possible to replace the components in the above-described embodiments with well-known components as appropriate, without departing from the spirit of this disclosure, and the above-described embodiments may also be combined as appropriate.

[0083] The contents described in each of the above embodiments can be understood, for example, as follows:

[0084] (1) A power plant control system according to one embodiment is a power plant control system for controlling a power plant capable of generating electricity by driving a turbine connected to a generator with steam generated by a boiler device, comprising: an operation data acquisition unit for acquiring operation data of the power plant; an optimization unit for determining optimized operation conditions by predicting process values ​​of the power plant corresponding to the operation data using a prediction model constructed by machine learning with the operation data as training data, and optimizing the operation conditions of the power plant based on the process values; a storage unit for storing the optimized operation conditions in association with the operating state of the power plant; and an operation condition setting unit for setting the operation conditions of the power plant, wherein the operation condition setting unit sets the past optimized operation conditions as operation conditions when it determines that past optimized operation conditions corresponding to the operating state of the power plant stored in the storage unit are applicable as operation conditions.

[0085] According to the embodiment of (1) above, as operating conditions for controlling the power plant, optimized operating conditions are set based on the process values ​​of the power plant predicted using a predictive model constructed with operating data corresponding to the operating state of the power plant as learning data. Therefore, in this embodiment, by setting past optimized operating conditions stored in the memory as operating conditions, the deterioration of plant performance in the changed operating state can be effectively suppressed.

[0086] (2) In another embodiment, in the embodiment of (1) above, if the operation condition setting unit determines that past optimization operation conditions corresponding to the operating state of the power plant stored in the memory unit are applicable as operation conditions, the unit sets the past optimization operation conditions as operation conditions while the optimization unit is calculating the optimization operation conditions.

[0087] According to the embodiment of (2) above, when the operating state of the power plant changes, calculating the optimization operating conditions using an optimization process with a predictive model for the changed operating state takes a considerable amount of time. However, by setting past optimization operating conditions stored in the memory unit as operating conditions during that period, the deterioration of the plant performance in the changed operating state can be effectively suppressed.

[0088] (3) In other embodiments, in the embodiment of (1) or (2) above, if the optimized operating conditions corresponding to the operating state do not exist in the storage unit, the operating condition setting unit sets the optimized operating conditions obtained by the optimization unit as the operating conditions.

[0089] According to the embodiment of (3) above, if no optimization operating conditions corresponding to the operating state of the power plant exist in the memory unit, the optimization unit can perform calculations again to suitably determine the optimization operating conditions corresponding to the new operating state.

[0090] (4) In other embodiments, in any one embodiment of (1) to (3) above, if the control state of the power plant is not normal when the past optimized operating conditions are set as the operating conditions, the operating condition setting unit sets a pre-prepared base operating condition as the operating conditions instead of the past optimized operating conditions.

[0091] According to the embodiment of (4) above, if the control state of the power plant is determined to be abnormal as a result of setting past optimized operating conditions as operating conditions, then, instead of the past optimized operating conditions, base operating conditions, which are pre-prepared as standard operating conditions corresponding to the operating state, are set as operating conditions. In this way, if the past optimized operating conditions are not suitable for the current operating state, the deterioration of plant performance can be effectively suppressed by setting base operating conditions as operating conditions.

[0092] (5) In another embodiment, in the embodiment of (4) above, if the operating parameters of the power plant, or the evaluation index calculated based on the operating parameters, deviate from an acceptable range, the control state is determined to be abnormal.

[0093] According to the embodiment of (5) above, whether or not the control state is normal when past optimization operating conditions are set as operating conditions can be determined by comparing the operating parameters of the power plant, or an evaluation index calculated based on the operating parameters, with an acceptable range.

[0094] (6) In other embodiments, in any one embodiment of (1) to (5) above, if a predetermined period has elapsed since the calculation of the past optimization operation conditions, it is determined that the past optimization operation conditions cannot be applied as the operation conditions.

[0095] According to the embodiment of (6) above, if a sufficiently long time has passed since the calculation of past optimization operating conditions, the past optimization operating conditions are deemed outdated and unsuitable for setting as operating conditions. This effectively prevents a decrease in plant performance due to the setting of unreliable, outdated past optimization operating conditions as operating conditions.

[0096] (7) In other embodiments, in any one embodiment of (1) to (6) above, if the operating conditions of the power plant have changed since the calculation of the past optimized operating conditions, it is determined that the past optimized operating conditions cannot be applied as the operating conditions.

[0097] According to the embodiment of (7) above, if the operating conditions of the power plant have changed since the calculation of past optimized operating conditions, those past optimized operating conditions are determined to be unsuitable for setting as operating conditions. This effectively prevents a decrease in plant performance due to past optimized operating conditions calculated under different operating conditions.

[0098] (8) In other embodiments, in any one embodiment of (1) to (7) above, if the past optimization operation conditions are not stored in the storage unit, the operation condition setting unit sets the past optimization operation conditions corresponding to the operating state predicted based on other past optimization operation conditions corresponding to other operating states as the operation conditions.

[0099] According to the embodiment of (8) above, even if past optimization operation conditions corresponding to a specific operating state are not stored in the memory unit, those past optimization operation conditions can be predicted based on past optimization operation conditions corresponding to other operating states.

[0100] (9) In other embodiments, in any one embodiment of (1) to (8) above, the optimization unit calculates a first optimization operation condition and a second optimization operation condition which has a greater likelihood than the first optimization operation condition for the allowable range of the process value as the optimization operation conditions corresponding to the operating state, and the operation condition setting unit sets the first optimization operation condition as the operation condition when the operating state is in a stable state, and sets the second optimization operation condition as the operation condition when the operating state is in a transient state.

[0101] According to the embodiment described in (9) above, the first and second optimized operating conditions are calculated by an optimization process using process values ​​predicted using a prediction model for the operating state of the power plant. The optimization operating conditions used to calculate the second optimized operating conditions are set to have a higher likelihood of the process values ​​being within the acceptable range compared to the optimization operating conditions used to calculate the first optimized operating conditions. As a result, the second optimized operating conditions are calculated as optimized operating conditions with a higher likelihood of the process values ​​being within the acceptable range compared to the first optimized operating conditions. The first optimized operating conditions are set as operating conditions when the operating state is in a stable state, and the second optimized operating conditions are set as operating conditions when the operating state is in a transient state. In this way, controllability can be improved by setting the second optimized operating conditions, which have a high likelihood, as operating conditions in transient states where process values ​​and indicators calculated from process values ​​are likely to deviate from the acceptable range. On the other hand, in stable states where process values ​​and indicators calculated from process values ​​are unlikely to deviate from the acceptable range, economic efficiency can be improved by setting the first optimized operating conditions, which have a low likelihood, as operating conditions.

[0102] (10) In other embodiments, in any one embodiment of (1) to (9) above, the optimization unit calculates the optimization operating conditions for each load zone of the boiler system as the operating state.

[0103] According to the embodiment of (10) above, the optimization operating conditions are calculated for each load range of the boiler system, which reduces the computational load compared to when they are calculated for each load, while improving control stability when the operating conditions change.

[0104] (11) In other embodiments, in any one embodiment of (1) to (10) above, the operating state is defined by a parameter including at least one of the type of fuel supplied to the boiler system or the number of operating fuel supply devices for supplying the fuel.

[0105] According to the embodiment of (11) above, the operating state of the power plant is preferably defined by a parameter that includes at least one of the type of fuel supplied to the boiler equipment or the number of operating fuel supply devices for supplying fuel (for example, a mill for crushing fuel, or a source of liquid or gaseous fuel).

[0106] Furthermore, the generator may be electrically connected to a power grid supplied with electricity from renewable energy power generation equipment.

[0107] In this case, the power plant being controlled is one that has a generator connected to a common power grid with renewable energy power generation equipment, the amount of power generated being greatly affected by environmental conditions. In this case, when the amount of power generated by the renewable energy power generation equipment changes, the power supply command to the power plant changes, and consequently, the operating state of the power plant changes. Even when the operating state changes in this way, as described in each of the embodiments above, when the operating state of the power plant changes, while the optimization process using the predictive model is performed for the changed operating state, past optimization operation conditions stored in the memory unit can be set as operating conditions, thereby effectively suppressing the deterioration of the plant performance in the changed operating state.

[0108] (12) A power plant control method according to one embodiment is a power plant control method for controlling a power plant capable of generating electricity by driving a turbine connected to a generator with steam generated by a boiler device, comprising: a step of acquiring operating data of the power plant; a step of predicting process values ​​of the power plant corresponding to the operating data using a predictive model constructed by machine learning with the operating data as training data, and determining optimized operating conditions by optimizing the operating conditions of the power plant based on the process values; a step of storing the optimized operating conditions in a memory unit in association with the operating state of the power plant; and a step of setting the operating conditions of the power plant, wherein in the step of setting the operating conditions, if it is determined that past optimized operating conditions corresponding to the operating state of the power plant stored in the memory unit are applicable as operating conditions, the past optimized operating conditions are set as operating conditions.

[0109] According to the embodiment of (12) above, as operating conditions for controlling the power plant, optimized operating conditions are set based on the process values ​​of the power plant predicted using a predictive model constructed with operating data corresponding to the operating state of the power plant as learning data. Therefore, in this embodiment, by setting past optimized operating conditions stored in the memory as operating conditions, the deterioration of plant performance in the changed operating state can be effectively suppressed.

[0110] (13) A power plant control program according to one embodiment is a power plant control program for controlling a power plant capable of generating electricity by driving a turbine connected to a generator with steam generated by a boiler device, wherein a computer device is capable of performing the following steps: acquiring operating data of the power plant; predicting process values ​​of the power plant corresponding to the operating data using a prediction model constructed by machine learning with the operating data as training data, and determining optimized operating conditions by optimizing the operating conditions of the power plant based on the process values; storing the optimized operating conditions in a memory unit in association with the operating state of the power plant; and setting the operating conditions of the power plant, wherein in the step of setting the operating conditions, if it is determined that past optimized operating conditions corresponding to the operating state of the power plant stored in the memory unit are applicable as the operating conditions, the past optimized operating conditions are set as the operating conditions.

[0111] According to the embodiment of (13) above, as operating conditions for controlling the power plant, optimized operating conditions are set based on the process values ​​of the power plant predicted using a predictive model constructed with operating data corresponding to the operating state of the power plant as learning data. Therefore, in this embodiment, by setting past optimized operating conditions stored in the memory as operating conditions, the deterioration of plant performance in the changed operating state can be effectively suppressed.

[0112] 1 Power plant control system 2 Power plant 4 Boiler 6 Steam turbine 8 Generator 10 Fuel supply system 12 Air supply system 16 Condenser 18 Feedwater pump 20 Fuel temperature sensor 22 Air temperature sensor 24 Steam temperature sensor 26 Steam pressure sensor 28 Condensate temperature sensor 100 Control device 200 Control support device 202 Operation data acquisition unit 204 Prediction model construction unit 206 Optimization unit 208 Storage unit 210 Operation condition setting unit

Claims

1. A power plant control system for controlling a power plant capable of generating electricity by driving a turbine connected to a generator with steam generated by a boiler system, comprising: an operation data acquisition unit for acquiring operation data of the power plant; an optimization unit for determining optimized operation conditions by predicting process values ​​of the power plant corresponding to the operation data using a predictive model constructed by machine learning with the operation data as training data, and optimizing the operation conditions of the power plant based on the process values; a storage unit for storing the optimized operation conditions in association with the operating state of the power plant; and an operation condition setting unit for setting the operation conditions of the power plant, wherein the operation condition setting unit sets the past optimized operation conditions as the operation conditions when it is determined that past optimized operation conditions corresponding to the operating state of the power plant stored in the storage unit are applicable as operation conditions.

2. The power plant control system according to claim 1, wherein the operation condition setting unit determines that past optimization operation conditions corresponding to the operating state of the power plant stored in the memory unit are applicable as operation conditions, and sets the past optimization operation conditions as operation conditions while the optimization unit is calculating the optimization operation conditions.

3. The power plant control system according to claim 1 or 2, wherein if the operating condition setting unit does not have the optimized operating condition corresponding to the operating state in the storage unit, the optimized operating condition determined by the optimization unit is set as the operating condition.

4. The power plant control system according to claim 1 or 2, wherein if the control state of the power plant is not normal when the past optimized operating conditions are set as the operating conditions, the operating condition setting unit sets pre-prepared base operating conditions as the operating conditions instead of the past optimized operating conditions.

5. The power plant control system according to claim 4, wherein if the operating parameters of the power plant, or the evaluation index calculated based on the operating parameters, deviate from an acceptable range, the control state is determined to be abnormal.

6. The power plant control system according to claim 1 or 2, which determines that the past optimization operating conditions cannot be applied as operating conditions if a predetermined period has elapsed since the calculation of the past optimization operating conditions.

7. The power plant control system according to claim 1 or 2, which determines that the past optimized operating conditions cannot be applied as operating conditions if the operating conditions of the power plant have changed since the calculation of the past optimized operating conditions.

8. The power plant control system according to claim 1 or 2, wherein if the past optimization operating conditions are not stored in the memory unit, the operating condition setting unit sets the past optimization operating conditions corresponding to the operating state predicted based on other past optimization operating conditions corresponding to other operating states as the operating conditions.

9. The power plant control system according to claim 1 or 2, wherein the optimization unit calculates a first optimization operation condition and a second optimization operation condition which has a greater likelihood than the first optimization operation condition for the allowable range of the process value as optimization operation conditions corresponding to the operating state, and the operation condition setting unit sets the first optimization operation condition as the operation condition when the operating state is in a stable state, and sets the second optimization operation condition as the operation condition when the operating state is in a transient state.

10. The power plant control system according to claim 1 or 2, wherein the optimization unit calculates the optimization operating conditions for each load zone of the boiler equipment as the operating state.

11. The power plant control system according to claim 1 or 2, wherein the operating state is defined by a parameter including at least one of the type of fuel supplied to the boiler system or the number of operating fuel supply devices for supplying the fuel.

12. A power plant control method for controlling a power plant capable of generating electricity by driving a turbine connected to a generator with steam generated by a boiler device, comprising: a step of acquiring operating data of the power plant; a step of predicting process values ​​of the power plant corresponding to the operating data using a predictive model constructed by machine learning with the operating data as training data, and determining optimized operating conditions by optimizing the operating conditions of the power plant based on the process values; a step of storing the optimized operating conditions in a memory unit in association with the operating state of the power plant; and a step of setting the operating conditions of the power plant, wherein in the step of setting the operating conditions, if it is determined that past optimized operating conditions corresponding to the operating state of the power plant stored in the memory unit are applicable as the operating conditions, the past optimized operating conditions are set as the operating conditions.

13. A power plant control program for controlling a power plant capable of generating electricity by driving a turbine connected to a generator with steam generated by a boiler device, wherein a computer device is capable of performing the following steps: acquiring operating data of the power plant; predicting process values ​​of the power plant corresponding to the operating data using a predictive model constructed by machine learning with the operating data as training data, and determining optimized operating conditions by optimizing the operating conditions of the power plant based on the process values; storing the optimized operating conditions in a memory unit in association with the operating state of the power plant; and setting the operating conditions of the power plant, wherein in the step of setting the operating conditions, if it is determined that past optimized operating conditions corresponding to the operating state of the power plant stored in the memory unit are applicable as the operating conditions, the past optimized operating conditions are set as the operating conditions.

Images