JP4423617B2 - Plant control device - Google Patents

Description

  The present invention relates to a plant control device such as a thermal power plant.

  In the plant control apparatus, a measurement signal obtained from a plant to be controlled is processed, and an operation signal to be given to the control target is calculated. The control device is implemented with an algorithm for calculating an operation signal so that the measurement signal of the plant satisfies the target value.

  As a control algorithm used for plant control, there is a PI (proportional / integral) control algorithm. In PI control, a value obtained by integrating the deviation with time is added to a value obtained by multiplying the deviation between the measurement signal of the plant and its target value by a proportional gain to derive an operation signal to be given to the controlled object. A control algorithm using PI control can be visualized by a block diagram or the like, and its operation principle and reason can be clearly understood. In general, such a block diagram is created by a control logic designer and does not perform an operation not intended by the designer. PI control is a stable and safe control algorithm and has many achievements.

  However, when the plant is operated under conditions that are not anticipated in advance, such as a change in the plant operation mode or a change in the environment, an operation such as changing the control logic is required.

  On the other hand, since the control algorithm is automatically corrected and changed in accordance with changes in the operation mode or environment of the plant, the plant can be controlled using adaptive control or a learning algorithm. As an example of a method for controlling a plant using a learning algorithm, a technique related to a control device using a reinforcement learning method has been proposed (see, for example, Patent Document 1). This control apparatus includes a model for predicting the characteristics of the controlled object, and learning means for learning a model input generation method that achieves the target value of the model output.

Japanese Unexamined Patent Publication No. 2000-35956 (pages 3-4)

  If the technique described in Patent Document 1 is used, the control algorithm can be automatically corrected / changed in accordance with changes in the operation mode and environment of the plant.

  However, in order to evaluate the validity of the learning results, it is necessary for humans to analyze the learning results in detail. Further, in order to operate the plant safely, it is necessary to confirm that the modified / modified algorithm operates normally.

  In other words, it is essential to improve the reliability of the learning algorithm in order to safely use the learning algorithm that can adapt to changes in the operation mode and environment of the plant for plant control.

  The subject of this invention is providing the plant control apparatus provided with the means by which the operator of a plant can confirm a learning result easily, in order to improve the reliability of a learning algorithm.

In order to solve the above-described problem, the present invention provides a plant control apparatus having an operation signal generation unit that calculates an operation signal to be given to a plant using a measurement signal of an operation state quantity of the plant. A measurement signal database for storing measurement signals, an operation signal database for storing past operation signals, numerical analysis execution means for analyzing plant operating characteristics, and numerical analysis results obtained by operating the numerical analysis execution means A numerical analysis result database to be stored, a model for estimating the value of a measurement signal when an operation signal is given to the plant using the information of the numerical analysis result database, and learning for learning a plant operation method using the model Means, a learning information database for storing learning information obtained by the learning means, and an operation signal generating means. A control logic database for storing information used in the derivation of the signal, and analyzing means for processing the information of the numerical analysis result database using the learning information database and the operation signal database information of the measurement signal database, An analysis result database for storing results analyzed by the analysis means, wherein the analysis means transmits a numerical analysis result for evaluating the validity of the operation method learned by the learning means to the analysis result database and rationale analysis means, a function of transmitting the numerical analysis results to assess the effect of giving an operation signal to the plant to the analysis result database, abnormal numerical analysis to evaluate the existence of the measurement signals results and signal analysis hand stage having at least one function of the ability to send measurement signals to said analysis result database Even without including one proposes a plant control system which comprises said analysis result display can display the data in the database.

According to the present invention, it can display the analysis results of the learning result on the screen. Thereby, the operator of a plant can confirm the analysis result of a learning result. Furthermore, since an operator of the plant can determine whether or not to input the operation signal derived by the learning algorithm to the plant, the reliability of the control device that implements the learning algorithm is increased.

  Moreover, when the measurement signal from a plant becomes an abnormal value, the factor can also be estimated. When it is necessary to modify or repair the plant body in order to avoid an abnormal state, it is also possible to display the contents of the modification or repair and the effect of the modification or repair on the screen.

  Furthermore, the control device of the present invention can be utilized as a training simulator for plant operators, and can be used to improve the skills of operators.

  Next, with reference to FIGS. 1-31, the Example of the plant control apparatus by this invention is described.

  FIG. 1 is a block diagram showing a system configuration of Embodiment 1 of a plant control apparatus according to the present invention. In FIG. 1, the plant 100 is controlled by a control device 200.

  The control device 200 that controls the plant 100 to be controlled is provided with a numerical analysis execution unit 220, a model 230, a learning unit 260, an operation signal generation unit 280, and an analysis unit 300 as arithmetic units.

  Further, the control device 200 is provided with a measurement signal database 210, a numerical analysis result database 240, an operation signal database 250, a learning information database 270, a control logic database 290, a knowledge database 400, and an analysis result database 500 as databases. .

  In addition, the control device 200 is provided with an external input interface 201 and an external output interface 202 as interfaces with the outside.

  In the control device 200, the measurement signal 1 is taken into the control device 200 from the plant 100 via the external input interface 201. In addition, the operation signal 15 is sent to the plant 100 via the external output interface.

  In the control device 200, the measurement signal 1 of the plant 100 is captured via the external input interface 201, and the captured measurement signal 2 is stored in the measurement signal database 210. Further, the operation signal 13 generated by the operation signal generation means 280 is transmitted to the external output interface 202 and stored in the operation signal database 250.

  The operation signal generation unit 280 generates the operation signal 13 using the control logic data 12 of the control logic database 290 and the learning information data 11 of the learning information database 270 so that the measurement signal 1 achieves the driving target value. The control logic database 290 stores a control circuit and control parameters for calculating the control logic data 12 in order to output the control logic data 12.

  Learning data stored in the learning information database 270 is generated by the learning means 260. The learning unit 260 is connected to the model 230.

  The model 230 has a function of simulating the control characteristics of the plant 100. That is, the operation signal 15 which becomes a control command is given to the plant 100, and the same operation as obtaining the measurement signal 1 of the control result is simulated. For this simulation calculation, the model input 7 for operating the model 230 is received from the learning means 260, and the control operation of the plant 100 is simulated by the model 230, and the model output 8 of the simulation calculation result is obtained. Here, the model output 8 is a predicted value of the measurement signal 1 of the plant 100.

  This model 230 is constructed based on the numerical analysis results in the numerical analysis result database 240. The numerical analysis result stored in the numerical analysis result database 240 is generated by the numerical analysis execution means 220.

  The numerical analysis execution unit 220 predicts the characteristics of the plant 100 using a physical model that simulates the plant 100. The numerical analysis execution means 220 executes calculation using, for example, a three-dimensional analysis tool and predicts the characteristics of the plant 100 in three dimensions. Calculation results obtained by the numerical analysis executing means 220 are stored in the numerical analysis result database 240.

  In the model 230, the model output 8 corresponding to the model input 7 is calculated using a statistical technique such as a neural network using the information of the numerical analysis result database 240 and the measurement signal database 210. The model 230 extracts a numerical analysis result necessary for calculating the model output 8 corresponding to the model input 7 from the numerical analysis result database 240 and interpolates the result. Further, in preparation for the difference between the model characteristics of the numerical analysis execution means 220 and the characteristics of the plant 100, the model 230 and the characteristics of the plant 100 are matched using the measurement signal data 3 of the measurement signal 210. Can be corrected.

  The learning means 260 learns how to generate the model input 7 so that the model output 8 simulated by the model 230 achieves the model output target value set in advance by the operator. The learning means 260 can be constructed by applying various optimization methods such as reinforcement learning and evolutionary calculation methods.

  Learning information data 9 such as constraint conditions and model output target values used for learning is stored in the learning information database 270. Further, the learning information data 10 that is the result of learning by the learning means 260 is stored in the learning information database 270. The learning information data 10 includes information on the state before the model input change and the method for changing the model input in this state.

  The analysis means 300 is connected to the measurement signal database 210, the operation signal database 250, the numerical analysis result database 240, the learning information database 270, the knowledge database 400, and the analysis result database 500, and extracts information from these databases. It has a function to execute various analyzes using the extracted information and send the analysis results to the database. The knowledge database 400 stores knowledge about the plant 100. For example, a phenomenon that occurs when the measured value of the plant 100 is greater than or equal to a certain value, knowledge about the causal relationship between the performance of the plant 100 and the measured value, and the like are stored. The detailed function of the analysis unit 300 will be described later.

  An operator of the plant 100 uses the external input device 600 including the keyboard 601 and the mouse 602, the maintenance tool 610 provided with the data transmission / reception means 630 capable of transmitting / receiving data to / from the control device 200, and the image display device 650 to perform control. Information of various databases provided in the apparatus 200 can be accessed.

  The maintenance tool 610 includes an external input interface 620, data transmission / reception means 630, and an external output interface 640.

  The maintenance tool input signal 51 generated by the external input device 600 is taken into the maintenance tool 610 via the external input interface 620. The data transmission / reception means 630 of the maintenance tool 610 acquires the database information 50 provided in the control device 200 according to the information of the maintenance tool input signal 52.

  The data transmission / reception processing means 630 transmits a maintenance tool output signal 53 obtained as a result of processing the database information 50 to the external output interface 640. The maintenance tool output signal 34 is displayed on the image display device 650.

  In the control device 200 of the above embodiment, the measurement signal database 210, the numerical analysis result database 240, the operation signal database 250, the learning information database 270, the control logic database 290, the knowledge database 400, and the analysis result database 500 are included in the control device 200. However, it is also possible to arrange all or one of these means outside the control device 200.

  Further, although the model 230, the learning unit 260, and the analysis unit 300 are arranged inside the control device 200, all or one of them can be arranged outside the control device 200.

  For example, the learning unit 260, the model 230, the numerical analysis execution unit 220, the numerical analysis result database 240, and the learning information database 270 are arranged outside the control device 200, and the numerical analysis result 16, the learning information data 11, and the learning information data 17 may be transmitted to the control device 200 via the Internet.

  FIG. 2 is a system diagram showing the analysis means 300. The analysis unit 300 includes a learning basis analysis unit 310, a signal analysis unit 320, and a knowledge database update unit 330.

  The learning basis analysis unit 310 analyzes the reason why the learning result of the learning information database 270 is obtained while referring to the information of the numerical analysis result database 240 and the knowledge database 400, and transmits the analysis result to the analysis result database 500. .

  The signal analysis means 320 processes the information in the operation signal database 250 and the measurement signal database 210 while referring to the information in the numerical analysis result database 240 and the knowledge database 400, and the effect of updating the operation signal and the abnormal measurement signal The analysis result is transmitted to the analysis result database 500.

  The knowledge database updating unit 330 evaluates the knowledge of the knowledge database 400 by referring to information in the numerical analysis result database 240, the measurement signal database 210, and the analysis result database 500. Further, it has a function of correcting the knowledge in the knowledge database 400 as needed, or a function of adding new knowledge.

  FIG. 3 is a flowchart showing the operation of the learning basis analysis unit 310. The operation of the learning basis analysis unit 310 is executed by combining Steps 2000, 2010, 2020, 2030, 2040, 2050, 2060, and 2070.

  In step 2000, I, which is a value indicating the number of repetitions of steps 2010 to 2060, is initialized (I = 1 is set). Next, in step 2010, initial values of operation conditions are set. In step 2020, the model input change width in the operation condition set in step 2010 is acquired from the learning information database 270.

  In step 2030, two numerical analysis results corresponding to the operation condition set in step 2010 and the operation condition after adding the model input change width derived in step 2020 to the model input of this operation condition are numerically analyzed. Extract from the result database 240.

  Next, in Step 2040, it is evaluated whether or not the knowledge of the knowledge database 400 can explain the difference between the two numerical analysis results extracted in Step 2030. Step 2050 is a branch, and if it is determined that it can be described in Step 2040, the process proceeds to Step 2060, and if it is determined that it cannot be described, the process proceeds to Step 2070.

  In step 2060, the two operation conditions, the numerical analysis result, and the knowledge used in step 2040 are transmitted to the analysis result database 500, and these pieces of information are stored in the analysis result database.

  In Step 2070, I, which is a value indicating the number of repetitions of Steps 2010 to 2060, is compared with a predetermined threshold value. If I is smaller than the threshold value, 1 is added to I, and the process returns to Step 2010. On the contrary, when the value is larger than the threshold value, the process proceeds to the step of ending the operation of the learning basis analysis unit 310.

  By the above operation, the analysis result database 500 stores the numerical analysis result before and after the model input update and the knowledge that can explain the difference between the two numerical analysis results. An operator of the plant 100 can display such information on the image display device 650 using the maintenance tool 610. Thereby, it can be easily recognized whether the operation method learned by the learning means 260 is appropriate. It is also possible to know the effect of changing the model input.

By the above operation, the analysis result database 500 stores the numerical analysis result before and after the model input update and the knowledge that can explain the difference between the two numerical analysis results. Operator of the plant 100 uses the maintenance tool 610, that Ki out to display the information to the image display device 650. Thereby, it can be easily recognized whether the operation method learned by the learning means 260 is appropriate. It is also possible to know the effect of changing the model input.

  The operation signal evaluation function block executes a combination of steps 2100, 2110, 2120, 2130, and 2140, and the measurement signal evaluation function block executes a combination of steps 2210, 2220, 2230, 2240, 2250, and 2260.

  The operation signal evaluation function block derives model inputs corresponding to before and after changing the operation signal, analyzes numerical analysis results corresponding to these two model inputs, and refers to the knowledge in the knowledge database 400 to thereby obtain the operation signal. The reason why the control characteristics change with the change can be considered.

  In step 2110, operation signals before and after changing the operation signal are extracted from the operation signal database 250, and two model inputs are derived. In step 2120, the numerical analysis results corresponding to the two model inputs are extracted from the numerical analysis result database 240. In step 2130, knowledge explaining the difference between the two numerical analysis results extracted in step 2120 is extracted from the knowledge database 400. Finally, the model input derived in step 2110, the numerical analysis result extracted in step 2120, and the knowledge extracted in step 2130 are transmitted to the analysis result database 500, and these pieces of information are stored in the analysis result database 500.

  In the measurement signal evaluation function block, the measurement signal is compared with the numerical analysis result, and when this difference is large, it is determined that the measurement signal of the plant 100 is an abnormal value, and the abnormality is detected using the knowledge of the knowledge database 400. The factor that is the value can be estimated.

  In step 2210, the current operation condition is extracted from the operation signal database 250, and a model input corresponding to this is derived. In step 2220, a numerical analysis result corresponding to the model input derived in step 2210 is extracted. In step 2230, the deviation between the numerical analysis result and the measurement signal is calculated. In step 2240, if the deviation calculated in step 2230 is larger than a predetermined threshold value, the process proceeds to step 2250, and if smaller, the process ends. In step 2250, the knowledge of the knowledge database 400 is utilized to analyze the reason why the deviation is larger than the threshold value. In step 2260, the analysis result in step 2250 is transmitted and stored in the analysis result database 500.

  FIG. 5 is a flowchart showing the operation of the knowledge database update unit 330. The knowledge database update unit 330 has three functional blocks: a knowledge evaluation function block, a knowledge correction function block, and a knowledge addition function block. In step 2300, which functional block to use is selected from the three functional blocks.

  The knowledge evaluation function block executes steps 2310, 2320, and 2330 in combination, the knowledge correction function block executes steps 2410, 2420, and 2430 in combination, and the knowledge addition function block includes steps 2510, 2520, and 2530. , 2540, and 2550 are executed in combination.

  In the knowledge evaluation function block, the effectiveness of knowledge in the knowledge database 400 is evaluated.

  In step 2310, knowledge used by the learning basis analysis unit 310 and the signal analysis unit 320 is extracted from the analysis result database 500, and the use count of each knowledge in the knowledge database 400 is derived. In step 2320, an evaluation value is calculated for each knowledge based on the number of uses derived in step 2310. For example, the evaluation value of knowledge having a large number of uses is set to a large value, and the legal value of knowledge having a low number of uses is set to a small value. In step 2330, the evaluation value calculated in step 2320 is transmitted and stored in the knowledge database 400.

  When the learning basis analysis unit 310 and the signal analysis unit 320 are executed, the knowledge in the knowledge database 400 is referred to in steps 2040, 2130, and 2250. Although all the knowledge in the knowledge database 400 can be referred to, if the number of knowledge is large, there is a possibility that the calculation processing time becomes long. In order to avoid this, when referring to knowledge in the knowledge database 400, if some knowledge is selected from knowledge having a large evaluation value and only this knowledge is referred to, the knowledge can be referred to efficiently within a limited time.

  In the knowledge correction function block, information in the measurement signal database 210, the operation signal database 250, and the numerical analysis result database 240 is compared with the knowledge in the knowledge database 400, and the knowledge in the knowledge database 400, the measurement signal database 210, and the operation signal database are compared. 250 and the information in the numerical analysis result database 240 are inconsistent, it is determined that there is an error in knowledge, and the knowledge in the knowledge database 400 is corrected.

  In step 2410, knowledge to be corrected is selected based on the above-described determination. In step 2420, the knowledge is corrected so that the contradiction is resolved. In step 2430, the new knowledge corrected in step 2420 is transmitted and stored in the knowledge database 400.

  The knowledge correction function block can also be operated so as to correct knowledge having a low evaluation value.

  Finally, the knowledge adding function block will be described. In the knowledge addition function block, a hypothesis that can explain one numerical analysis result in the numerical analysis result database 240 is generated. If this hypothesis can sufficiently explain other numerical analysis results, this hypothesis is registered in the knowledge database 400 as knowledge.

  In step 2510, the numerical analysis result of the numerical analysis result database 240 is extracted. In step 2520, a hypothesis that can explain the numerical analysis result extracted in step 2510 is generated. In step 2530, it is evaluated whether or not the hypothesis generated in step 2520 can explain another numerical analysis result of the numerical analysis result database 240. In Step 2540, it is determined whether or not to add a hypothesis to the knowledge database 400 based on the evaluation result in Step 2530. When adding, it progresses to step 2550, and when not adding, a knowledge addition functional block is complete | finished. In step 2550, the hypothesis generated in step 2530 is transmitted and stored in the knowledge database 400.

  Above, description of the learning basis analysis means 310, the signal analysis means 320, and the knowledge database 330 which comprise the analysis means 300 is complete | finished.

  6 to 10 are examples of screens displayed on the image display device 650. FIG.

  FIG. 6 shows an initial screen. In a state where FIG. 6 is displayed on the image display device 650, the button can be selected by operating the mouse 602 so that the cursor is over the button and the mouse 602 is clicked. When the buttons 701, 702, 703, and 704 are selected, FIGS. 7, 8, 9, and 10 are displayed on the image display device 650.

  In FIG. 7, when the button 711 is selected, various analysis conditions necessary for executing the calculation by the numerical analysis execution means 220 can be input / set. When the button 712 is selected, the numerical analysis execution means 220 can start the calculation. When the button 713 is selected, the process returns to FIG.

  In FIG. 8, which database information is to be displayed on the image display device 650 can be selected. When the buttons 721, 722, 723, 724, 725, 726, 727 are selected, the measurement signal database 210, the operation signal database 250, the numerical analysis result database 240, the learning information database 270, the control logic database 290, the knowledge database 400, and the analysis, respectively. The result database 500 can be accessed. Information on each database can be displayed on the image display device 650, and information on the database can be added / changed / erased.

  In FIG. 9, when the button 731 is selected, the learning means 260 and the model 230 can be operated to learn the operation method. When the button 732 is selected, the learning basis analysis unit 310 can be operated.

  In FIG. 10, every time the operation signal for operating the plant 100 is changed, it is possible to select whether to display the effect of changing the operation signal on the image display device 650 by operating the signal analysis unit 320. The difference between the case where the button 741 is selected in FIG. 10 and the case where the button 742 is selected will be described with reference to FIG.

  FIG. 11 is a flowchart showing the operation of the control device 200. The control operation of the control device 200 is executed by combining steps 1010, 1020, 1030, 1040, 1050, and 1060.

  In step 1010, the operation signal generation means 280 is operated to generate a next operation signal candidate. Step 1020 is display determination. When the button 741 is selected in FIG. 10, the process proceeds to step 1040. When the button 742 is selected, the process proceeds to step 1030.

  In step 1040, the operation signal evaluation function block of the signal analysis means 320 is operated to estimate the effect of changing the operation signal. In step 1050, the operation effect is displayed on the image display device 650.

  FIG. 12 is an example of a screen displayed on the image display device 650 in step 1050. Two numerical analysis results corresponding to before and after the operation are displayed in 751 and 752, respectively. Knowledge explaining the difference between the two types of numerical analysis results is displayed in a column 753. The plant operator can select whether or not to perform the operation while viewing the information. When the operation is permitted, the button 754 is selected. When the operation is not permitted, the button 755 is selected.

  Step 1060 in FIG. 11 is an operation execution determination. If the button 754 in FIG. 12 is selected, the process proceeds to step 1030. If the button 755 is selected, the process returns to step 1010.

  In step 1030, the operation signal generated in step 1010 is given to the plant 100.

  Next, a second embodiment in which the plant control apparatus according to the present invention is applied to a thermal power plant will be described.

  Needless to say, the plant control apparatus of the present invention can also be used when controlling a plant other than a thermal power plant.

  FIG. 13 is a diagram showing a system configuration of the thermal power plant 100.

  The boiler 101 constituting the thermal power plant is provided with a burner 102 for supplying pulverized coal, which is fuel obtained by finely pulverizing coal in a mill 110, primary air for conveying pulverized coal, and secondary air for combustion adjustment. The pulverized coal supplied through the burner 102 is combusted inside the boiler 101. The pulverized coal and the primary air are led from the pipe 134 and the secondary air is led from the pipe 141 to the burner 102.

  Further, the boiler 101 is provided with an after air port 103 for introducing after-air for two-stage combustion into the boiler 101, and the after air is guided from the pipe 142 to the after-air port 103.

  The high-temperature combustion gas generated by the combustion of the pulverized coal flows downstream along the path inside the boiler 101, passes through the heat exchanger 106 disposed in the boiler 101, and performs heat exchange. Pass 104. The gas that has passed through the air heater 104 is subjected to exhaust gas treatment and then released from the chimney to the atmosphere.

  The feed water circulating through the heat exchanger 106 of the boiler 101 is supplied with the feed water to the heat exchanger 106 via the feed water pump 105, and is superheated by the combustion gas flowing down the boiler 101 in the heat exchanger 106. Become. In this embodiment, the number of heat exchangers is one, but a plurality of heat exchangers may be arranged.

  The high-temperature and high-pressure steam that has passed through the heat exchanger 106 is guided to the steam turbine 108 via the turbine governor 107, and the steam turbine 108 is driven by the energy of the steam to generate power by the generator 109.

  In the thermal power plant, various measuring devices that detect the operating state of the thermal power plant are arranged, and the measurement signal of the plant acquired from these measuring devices is transmitted to the control device 200 as the measurement signal 1. . For example, FIG. 2 illustrates a flow rate measuring device 150, a temperature measuring device 151, a pressure measuring device 152, a power generation output measuring device 153, and a concentration measuring device 154.

  The flow rate measuring device 150 measures the flow rate of the feed water supplied from the feed water pump 105 to the boiler 101. Further, the temperature measuring device 151 and the pressure measuring device 152 measure the temperature and pressure of the steam supplied from the heat exchanger 106 to the steam turbine 108.

  The amount of power generated by the power generator 109 is measured by a power generation output measuring device 153. Information on the concentration of components (CO, NOx, etc.) contained in the combustion gas passing through the boiler 101 can be measured by a concentration measuring device 154 provided on the downstream side of the boiler 101.

  In general, a number of measuring instruments other than those shown in FIG. 13 are arranged in the thermal power plant, but the illustration is omitted here.

  Next, the paths of primary air and secondary air that are input from the burner 102 into the boiler 101 and the path of after-air that is input from the after-air port 103 will be described.

  The primary air is led from the fan 120 to the pipe 130, and is branched into a pipe 132 that passes through the air heater 104 installed on the downstream side of the boiler 101 and a pipe 131 that bypasses without passing through the pipe, and is piped again. At 133, they join together and are guided to a mill 110 installed on the upstream side of the burner 102.

  The air passing through the air heater 104 is overheated by the combustion gas flowing down the boiler 101. Using this primary air, the differential charcoal crushed in the mill 110 is conveyed to the burner 102 together with the primary air.

  The secondary air and the after air are led from the fan 121 to the pipe 140 and heated in the same manner by the air heater 104, and then branched into the secondary air pipe 141 and the after air pipe 142, respectively. And the after-air port 103.

  FIG. 14 is a diagram showing a configuration of an air heater and piping in the thermal power plant shown in FIG. Air dampers 160, 161, 162, and 163 are disposed on the pipes 131, 132, 141, and 142, respectively. By operating these air dampers, the area through which the air passes through the pipes 131, 132, 141, 142 can be changed, so that the flow rate of air passing through the pipes 131, 132, 141, 142 can be individually adjusted.

  The operation signal 15 generated by the control device 200 is used to operate devices such as the water supply pump 105, the mill 110, and the air dampers 160, 161, 162, and 163. In the present embodiment, devices such as the water supply pump 105, the mill 110, and the air dampers 160, 161, 162, and 163 are referred to as operation ends, and command signals necessary for operating them are referred to as operation signals.

  In addition, when a fuel such as combustion or fuel such as pulverized coal is introduced into the boiler 101, a function capable of moving the discharge angle up and down and left and right is added to the burner 102 and the after air port 103. Can also be included in the operation signal 15.

  FIG. 15 is a diagram illustrating a mode of data stored in the measurement signal database 210 and the operation signal database 250. In the measurement signal database 210 and the operation signal database 250, values of each signal are stored in time series together with the unit. These pieces of information can be displayed on the image display device 650 by selecting the buttons 721 and 722 in FIG.

  FIG. 16 shows a screen for setting numerical analysis conditions in the numerical analysis execution means 220. The screen is displayed on the image display device 650 when the button 711 is selected in FIG. The operating conditions of the plant 100 are input using FIG. In FIG. 16, R_0001 is a number assigned to distinguish the analysis conditions, and FIG. 16 illustrates input means for five types of analysis conditions R_0001 to R_0005. In FIG. 16, the number of analysis conditions that can be input is five, but this number can be increased.

  When the analysis condition is input using FIG. 16 and the button 712 is selected in FIG. 7, the numerical analysis corresponding to the operation condition of FIG.

  FIG. 17 is a diagram illustrating an aspect of data stored in the numerical analysis result database 240. In the numerical analysis result database 240, the analysis conditions input in FIG. 16 and the numerical analysis results executed by the numerical analysis execution means 220 are stored in association with each other. Numerical analysis results can display items set by the operator. Further, by referring to the data described in the data file name column, it is possible to know the three-dimensional numerical analysis result. In FIG. 17, the CO concentration and the NOx concentration are displayed, but other than this, for example, the steam temperature, the gas temperature, and the like can also be displayed.

  FIG. 18 shows three-dimensional numerical analysis results stored in the numerical analysis result database 240. FIG. 18 shows the gas velocity vector, temperature, and CO concentration when cut in a cross section passing through the burner and after-air port in the third row among the burners and after-air ports arranged in five rows in the boiler 101 in the horizontal direction. , And NOx concentration.

  Further, the numerical analysis result on an arbitrary cut surface can be displayed on the image display device 650.

  FIG. 19 is a diagram illustrating the characteristics of the model 230 configured based on the numerical analysis results in the numerical analysis result database 240. The model 230 interpolates the numerical analysis results in the numerical analysis result database 240. Various methods such as a neural network can be used for the interpolation.

  When the control device 200 is applied to a thermal power plant and CO is to be reduced, the model 230 functions as a model for interpolating the numerical analysis result of CO. Since the calculation by the numerical analysis execution unit 220 is performed on a three-dimensional model, a long calculation time is required. In order to advance learning in a short time, it is necessary to acquire a numerical analysis result corresponding to the model input 7 in a short time.

  Therefore, in this embodiment, a numerical analysis result under a plurality of analysis conditions is stored in the numerical analysis result database 240, and a model for calculating the CO concentration in a short time is constructed by interpolating the numerical analysis results. Yes.

  FIG. 20 is a diagram illustrating an aspect of information stored in the learning information database 270. In the learning information database, the driving state and the value of the operation amount change width in the driving state are stored correspondingly. In FIG. 20, S_0001 is a number assigned to distinguish the operation state.

  FIG. 21 is a diagram illustrating an aspect of information stored in the knowledge database 400. The knowledge database 400 stores information in three types of formats: A type, B type, and C type.

  Type A includes the position of the gas passing through the boiler, and the gas temperature, gas flow rate, and gas composition at the position, such as nitrogen oxide, carbon monoxide, carbon dioxide, sulfide oxide, mercury, fluorine, Knowledge of the impact on plant performance factors, such as the amount or concentration of dust, mist particulates, or volatile organic compounds, or boiler operating efficiency is stored.

  In type B, knowledge about the effect obtained by changing the operation signal is stored.

  In Type C, an estimation factor that increases an error between the measurement signal and the numerical analysis result is stored as an analysis condition for confirming this. If the error is large, the plant may be operating in an abnormal state. Measures implemented to recover from this abnormal state and analysis conditions for performing numerical analysis assuming that the measures have been implemented are also stored.

  FIG. 22 is a diagram illustrating an aspect of information stored in the analysis result database 500. The analysis result database 500 stores the operation content, effect prediction, improvement probability, estimation factor, reference diagram, and past results.

  FIG. 23 is a flowchart showing the details of step 2040 when the control device of the present invention is used in a thermal power plant in FIG. This operation is executed by combining Steps 1110, 1120, and 1130.

  In step 1110, from the two types of numerical analysis results extracted in step 2030, gas concentration distributions such as CO and NOx in the boiler outlet cross section are compared, and a region in which the concentration deviation is equal to or greater than a threshold value is extracted. In step 1120, the gas path reaching the region extracted in step 1110 is traced in the direction opposite to the gas flow, and information relating to the gas temperature and gas flow velocity in this gas path is extracted. In step 1130, the knowledge of type A in the knowledge database 400 is compared with the extracted gas concentration / gas temperature / gas flow rate, and it is evaluated whether or not the content of the knowledge in the knowledge database 400 matches the numerical analysis result.

  FIG. 24 is a flowchart showing details of step 2130 when the control device of the present invention is used in a thermal power plant in FIG. This operation is executed by combining steps 1210 and 1220.

  In step 1210, the difference between the two model inputs extracted in steps 2110 and 2120 and the two model outputs corresponding to the model inputs are extracted. Next, in step 1220, knowledge that can explain the difference extracted in step 1210 is extracted from type B knowledge in the knowledge database 400.

  FIG. 25 is a flowchart showing the details of step 2250 when the control device of the present invention is used in a thermal power plant in FIG. 4 showing the operation of the signal analysis means 320. This operation is performed by combining Steps 1310, 1320, and 1330.

  In step 1310, the reason why the deviation is large is extracted from the items that are equal to or larger than the threshold value in step 2240 using the knowledge of type C in the knowledge database 400. In step 1320, a simulation for confirmation is executed. In step 1330, the numerical analysis result obtained as a result of the execution in step 1320 is compared with the measurement signal, and a simulation condition and a factor corresponding to the deviation are extracted.

  FIG. 26 is a flowchart showing the details of steps 2410 and 2420 when the control device of the present invention is used in a thermal power plant in FIG. This operation is performed by combining steps 1400, 1410, 1420, 1430, and 1440.

  In this embodiment, there are two methods for selecting knowledge to be corrected in step 2410, and in step 1400, which one is used is selected. In some cases, the knowledge to be corrected is selected in steps 1410 and 1420, and the knowledge to be corrected is selected in step 1430.

  In step 1410, the information of the measurement signal database 210, the operation signal database 250, and the numerical analysis result database 240 is compared with the knowledge of type A in the knowledge database 400. In step 1420, if there is a contradiction as a result of the comparison in step 1410, the process proceeds to step 1440. If there is no contradiction, it is determined that it is not necessary to correct the knowledge, and this operation is terminated. In step 1430, knowledge having a low evaluation value is extracted from knowledge in the knowledge database 400 and selected as knowledge to be corrected. In step 1440, the knowledge is corrected so that the information in the measurement signal database 210, the operation signal database 250, and the numerical analysis result database 240 does not conflict with the knowledge of the type A in the knowledge database 400. Specifically, the quantitative value of the information in the knowledge database 240 is corrected to a value that can explain all the values in the measurement signal database 210, the operation signal database 250, and the numerical analysis result database 240.

  FIG. 27 is a flowchart showing the details of steps 2520 and 2530 when the control device of the present invention is used in a thermal power plant in FIG. 5 showing the operation of the knowledge database updating means 330. Step 2520 is performed by combining Steps 1500, 1510, 1520, 1530, and 1540, and Step 2530 is performed by combining Steps 1550, 1560, 1570, 1580, and 1590.

  Step 1500 of Step 2520 is a branch according to the number of numerical analysis results extracted in Step 2510. If the number of numerical analysis results extracted is one, the process proceeds to Step 1510. Otherwise, Step 1520 is performed. Proceed to In step 1510, a region where the gas composition concentration at the boiler outlet is maximum is extracted, and in step 1520, the boiler outlet gas composition concentration of the two numerical analysis results is compared, and the region where the concentration difference is equal to or greater than the threshold value. To extract. In step 1530, a gas path reaching the region extracted in step 1510 or step 1520 is extracted. In step 1540, a hypothesis in which the maximum value of the gas temperature and the gas flow velocity in the gas path extracted in step 1530 is associated with the concentration is generated. That is, hypotheses are generated in the format of type A and type B in FIG.

  In step 1550 of step 2530, I is initialized (I = 1). In step 1560, the I-th numerical analysis result is compared with the hypothesis generated in step 1540. In step 1570, if there is no contradiction between the hypothesis and the I-th numerical analysis result in step 1560, the process proceeds to step 1580, and if there is a contradiction, the process proceeds to step 1590. In step 1580, the hypothesis evaluation value is increased. In step 1590, when all numerical analysis results in the numerical analysis result database 240 are not referred to, 1 is added to I and the process returns to step 1560. When all numerical analysis results are referred to, this operation is terminated. Thereafter, in step 2540, if the hypothesis evaluation value is equal to or greater than the threshold, the process proceeds to step 2550.

  FIG. 28 is an example of a screen displayed on the image display device 650 when the control device of the present invention is used in a thermal power plant, and is a screen corresponding to FIG. In this way, the simulation results before and after the operation are displayed on one screen, and the effect of the operation can be clarified. The plant operator can determine whether or not the operation can be performed while viewing the screen.

  FIG. 29 is an example of a screen displayed on the image display device 650 when the control device of the present invention is used in a thermal power plant, and shows simulation results before and after plant remodeling. Thus, by using the control device of the present invention, the effect of the plant modification can be displayed on one screen, so that the plant operator can make a quick decision.

  Next, FIG. 30 and FIG. 30 show an operation example of the analysis means 300 when the control device of the present invention is applied to a thermal power plant and the learning means 260 learns the operation method for obtaining the carbon monoxide (CO) concentration. 31 will be used for explanation.

  FIG. 30 is a diagram illustrating the result of numerical analysis performed by the numerical analysis execution unit 220 for the boiler constituting the thermal power plant, and is a diagram for explaining the distribution of the boiler outlet CO concentration. The result of the numerical analysis is stored in the numerical analysis result database 240.

  From the information in the numerical analysis result database 240, the CO concentration distribution in the cross section of the boiler outlet can be extracted. In step 1110 in FIG. 23, steps 1510 and 1520 in FIG. 27, etc., the gas composition concentration at the boiler outlet is automatically extracted in the form shown in FIG. The after air ports 103 arranged in five rows in the horizontal direction in the boiler 101 are referred to as after air ports 103A, 103B, 103C, 103D, and 103E, and the A, B, C, D and E are defined.

  FIG. 31 is a diagram for explaining a state in which numerical analysis results before and after an operation are compared when the plant control apparatus according to the present invention is applied to a thermal power plant, and particularly a diagram for explaining the detailed operation of FIG. It is.

  FIG. 31A shows operating conditions (operation command values) corresponding to before and after changing the operation signal, and numerical analysis results corresponding to the operating conditions.

  The learning means 260 learns the operation method using the model having the form shown in FIG. In FIG. 19, there is one operation condition, but here, the model has five operation conditions. As a result, as shown in FIG. 31A, an operation method for increasing the air flow rate supplied from the after air port 103B and decreasing the air flow rate supplied from the after air ports 103D and 103E was learned. As a result, the CO concentration at the boiler outlet was changed from 150 ppm to 100 ppm.

  However, the conventional control device can learn how to change the flow rate of air supplied from the after-air port, but cannot know the physical reason why the CO concentration can be reduced by making such a change. .

  In the control apparatus of the present invention, the analysis means 300 can analyze the reason why the CO concentration is reduced. Here, the operation will be described along the flowchart of FIG.

  In step 1110, as shown in FIG. 31A, a region 800 having a large concentration change is extracted from the numerical analysis result of the boiler outlet CO concentration before and after the operation.

  Next, in step 1120, as shown in FIG. 31B, the gas velocity vector is traced in reverse from the region 800 extracted in step 1110, and the gas path 810 reaching the region 800 is extracted. As a result, as shown in FIG. 31B, it can be seen that the gas starting from the after-air port 103B reaches the region 800. Further, as shown in FIG. 31 (c), the O2 concentration in the gas path 810 is automatically extracted.

  When coal is burned, CO is generated because incomplete combustion occurs in a region where the O2 concentration is low. As shown in FIG. 31 (c), before the operation, the O2 concentration at the boiler outlet is low, and as a result, CO is generated as shown in FIG. 31 (a). Therefore, by increasing the amount of air introduced from the after-air port 103B, the phenomenon that the O2 concentration is lowered at the boiler outlet is avoided, and as a result, the CO concentration is reduced.

  Information in the format shown in FIG. 21 is stored in the knowledge database 400, and knowledge that CO is likely to be generated when the O2 concentration is low is stored. In step 1130 of FIG. 23, comparing the above knowledge of the knowledge database 400 with the numerical analysis result of FIG. 31C, it can be avoided that the O2 concentration decreases when the air flow rate of the after-airport 103B is increased. As a result, it is automatically understood that the CO concentration can be reduced.

  The analysis result is displayed on the image display device 650 in the format shown in FIG. An operator of the plant can determine whether or not the operation can be executed after confirming the numerical analysis result and the analysis result. If it is determined that the numerical analysis result and the analysis result are not reliable, it is possible to prevent such an operation input from being given to the plant. That is, not only learning in the control device 200 but also introducing a step in which the plant operator confirms the plant 100 can be operated more safely.

  As for the analysis result in the analysis means 300, the same result can be obtained even if a human analyzes the numerical analysis result or the learning result.

  However, during plant operation, it is necessary to instantaneously determine whether or not to change the operation signal, and there is no room for human analysis. Therefore, the analysis means 300 provided in the control device 200 of the present invention is an essential function for safely operating the plant.

  When analyzing the thermal power plant, the knowledge database 400 is not limited to the above-mentioned knowledge about CO, but also particulates composed of nitrogen oxide (NOx), carbon dioxide, sulfide oxide, mercury, fluorine, dust, mist, or volatilization. Knowledge of at least one amount or concentration of the volatile organic compound as well as plant operating efficiency is stored.

  NOx includes fuel NOx produced by reaction of nitrogen contained in coal and thermal NOx produced by reaction of nitrogen contained in air, and thermal NOx is generated particularly in a high temperature region. . When the operation signal is changed and NOx at the boiler outlet decreases, the reason why NOx can be reduced can be analyzed by the same method as described with reference to FIGS. That is, the gas temperature in the gas path reaching the region where the change in the NOx concentration at the boiler outlet is large is illustrated as shown in FIG. 31 (c). Suppose it was. This phenomenon means that, by executing the operation, the increase in gas temperature is suppressed, the generation of thermal NOx is suppressed, and the NOx at the boiler outlet can be reduced.

  Such an analysis can be automatically executed by using the plant control apparatus of the present invention.

It is a block diagram which shows the system | strain structure of Example 1 of the plant control apparatus by this invention. It is a block diagram which shows an example of a structure of the analysis means in the plant control apparatus by this invention. It is a flowchart which shows operation | movement of the learning basis analysis means in the plant control apparatus by this invention. It is a flowchart which shows operation | movement of the signal analysis means in the plant control apparatus by this invention. It is a flowchart which shows operation | movement of the knowledge database update means in the plant control apparatus by this invention. It is an initial screen displayed on the image display apparatus connected with the plant control apparatus by this invention. In the plant control apparatus by this invention, when performing a numerical analysis, it is a screen displayed on an image display apparatus. In the plant control apparatus according to the present invention, when referring to the database, the screen is displayed on the image display apparatus. In the plant control apparatus by this invention, when performing learning, it is a screen displayed on an image display apparatus. In the plant control apparatus by this invention, when selecting the presence or absence of displaying an operation effect, it is a screen displayed on an image display apparatus. It is a flowchart which shows operation | movement of the plant control apparatus by this invention. In the plant control apparatus by this invention, when determining whether operation execution is possible, it is a screen displayed on an image display apparatus. It is a figure which shows an example of the system configuration | structure of the thermal power plant which should apply this invention. It is a figure which shows the structure of the air heater and piping in a thermal power plant. It is a figure which shows the aspect of the information preserve | saved at the measurement signal and the operation signal database, when the plant control apparatus by this invention is applied to a thermal power plant. When the plant control apparatus by this invention is applied to a thermal power plant, it is a screen displayed on an image display apparatus, when inputting analysis conditions. It is a figure which shows the aspect of the information preserve | saved at the numerical analysis result database, when the plant control apparatus by this invention is applied to a thermal power plant. It is a figure which shows the aspect of the information preserve | saved at the numerical analysis result database, when the plant control apparatus by this invention is applied to a thermal power plant. It is a figure explaining the construction method of a model, when the plant control apparatus by this invention is applied to a thermal power plant. It is a figure which shows the aspect of the information preserve | saved in the learning information database, when the plant control apparatus by this invention is applied to a thermal power plant. It is a figure which shows the aspect of the information preserve | saved in the knowledge database, when the plant control apparatus by this invention is applied to a thermal power plant. It is a figure which shows the aspect of the information preserve | saved in the analysis result database, when the plant control apparatus by this invention is applied to a thermal power plant. It is a flowchart which shows operation | movement of one means of a learning basis analysis means, when the plant control apparatus by this invention is applied to a thermal power plant. It is a flowchart which shows operation | movement of one means of the operation signal evaluation function which comprises a signal analysis means, when the plant control apparatus by this invention is applied to a thermal power plant. It is a flowchart which shows operation | movement of one means of the measurement signal evaluation function which comprises a signal analysis means, when the plant control apparatus by this invention is applied to a thermal power plant. It is a flowchart which shows operation | movement of one means of the knowledge correction function which comprises a knowledge database update means, when the plant control apparatus by this invention is applied to a thermal power plant. It is a flowchart which shows operation | movement of one means of the knowledge addition function which comprises a knowledge database update means, when the plant control apparatus by this invention is applied to a thermal power plant. When the plant control apparatus by this invention is applied to a thermal power plant, it is a screen displayed on an image display apparatus, when determining whether operation execution is possible. When the plant control apparatus by this invention is applied to a thermal power plant, it is a screen displayed on an image display apparatus, when evaluating the effect of plant remodeling. It is a distribution map of boiler exit CO density | concentration in the numerical analysis result of the boiler which comprises a thermal power plant. When the plant control apparatus by this invention is applied to a thermal power plant, it is a figure explaining a mode that the numerical analysis result before operation and after operation are compared.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Plant 200 Control apparatus 201 External input interface 202 External output interface 210 Measurement signal database 220 Numerical analysis execution means 230 Model 240 Numerical analysis result database 250 Operation signal database 260 Learning means 270 Learning information database 280 Operation signal generation means 290 Control logic database 300 Analysis means 400 Knowledge database 500 Analysis result database 600 External input device 601 Keyboard 602 Mouse 610 Maintenance tool 620 External input interface 630 Data transmission / reception processing means 640 External output interface 650 Image display device

Claims (6)

In a plant control apparatus having an operation signal generating means for calculating an operation signal to be given to the plant using a measurement signal of the operation state quantity of the plant,
The plant control device is
A measurement signal database for storing past measurement signals;
An operation signal database for storing past operation signals;
Numerical analysis execution means for analyzing the operation characteristics of the plant;
A numerical analysis result database for storing numerical analysis results obtained by operating the numerical analysis execution means;
A model for estimating a value of a measurement signal when an operation signal is given to the plant using information of the numerical analysis result database;
Learning means for learning a plant operation method using the model;
A learning information database for storing learning information obtained by the learning means;
A control logic database for storing information used for deriving an operation signal by the operation signal generating means;
And analyzing means for processing the information of the numerical analysis result database using the learning information database and the operation signal database information of the measurement signal database,
An analysis result database for storing results analyzed by the analysis means;
The analysis means is
Learning basis analysis means for transmitting a numerical analysis result for evaluating the validity of the operation method learned by the learning means to the analysis result database;
A numerical analysis function of transmitting the result to the analysis result database effect for evaluating if given an operation signal to the plant, the numerical analysis results and measurement signal to evaluate the existence of abnormal measurement signal the analysis includes at least one of the signal analysis hand stage having at least one function of the ability to send to the database,
A plant control device comprising a display device capable of displaying data of the analysis result database . The plant control apparatus according to claim 1,
The learning basis analysis means
Means for setting an initial value of the plant operating conditions used for analysis in the learning basis means;
Means for extracting an operation signal change width in the operation condition from the learning information database;
Means for extracting from the numerical analysis result database two types of numerical analysis results corresponding to the operation conditions and the plant operating conditions after the change of the operation signal;
A plant control apparatus comprising: means for transmitting the two numerical analysis results to the analysis result database in order to compare the two numerical analysis results and determine whether or not the difference is appropriate. The plant control apparatus according to claim 1,
The signal analysis means comprises:
Having at least one functional block of an operation signal evaluation functional block and a measurement signal evaluation functional block;
The operation signal evaluation function block includes:
Means for extracting the operation signal before the change and the operation signal after the change from the operation signal database;
Means for extracting two numerical analysis results corresponding to the two operation signals before and after the change from the numerical analysis result database;
Comprising means for storing the extracted numerical analysis results in the analysis result database;
The measurement signal evaluation function block includes:
Means for extracting a current operation signal from the operation signal database;
A plant control apparatus comprising: means for extracting a numerical analysis result corresponding to an operation signal from the numerical analysis result database. In the plant control device according to any one of claims 1 to 3,
Characterized in that a function block to display information of the measurement signal database and the operation signal database and the control logic database and the learning information database and the numerical analysis result database and knowledge database and the analysis result database on the screen A plant control device. In the plant control device according to claim 4,
The operation signal to be input in the next operation and the analysis result for evaluating the validity of inputting the operation signal to the plant are displayed on the screen, and whether or not to input the operation signal to the plant is indicated to the plant operator. A plant control device comprising a menu for selection. The plant control apparatus according to claim 1,
The plant is a thermal power plant, and the numerical analysis result stored in the numerical analysis result database is at least one of a gas velocity vector, a temperature, a CO concentration, and a NOx concentration. apparatus.

Images