Intelligent thermal power generating unit signal adaptive control method and system

Abstract

The invention discloses an intelligent thermal power generating unit signal adaptive control method and system, and belongs to the field of thermal power generating unit control, and the system comprises an acquisition processing module, a monitoring diagnosis module, an analogue simulation module, a state prediction module, a control optimization module, a decision evaluation module, a fault tolerance module and an interactive visualization module. According to the method, the global optimization capability of the system can be improved, the fault tolerance and the recovery capability of the unit are improved, dynamic adjustment and autonomous decision making are realized, and the stability and the reliability of the unit in the long-time operation process are improved; according to the method, the adaptability of the system in uncertainty and complex environments can be improved, the full life cycle efficiency of the units can be improved, the long-term operation cost can be reduced, the continuous and efficient operation of the system can be ensured, meanwhile, the power generation capacity of the units can be automatically adjusted or the combined dispatching of the multiple units can be adjusted, and energy waste or equipment damage can be avoided.

Description

Technical Field

[0001] The present invention relates to the field of thermal power unit control, and particularly to an intelligent signal adaptive control method and system for thermal power units. Background Art

[0002] With the continuous growth of global energy demand, thermal power units, as a traditional power generation method, still occupy an important position worldwide. As a complex mechanical system, thermal power units involve the coordinated operation of multiple key components such as boilers, steam turbines, generators, and auxiliary equipment. The parameter changes during their operation are highly non-linear and time-varying. Therefore, how to improve the operation stability, response ability, and fault tolerance of thermal power units has become a major challenge in the current power industry. Most traditional thermal power unit control systems adopt control algorithms based on experience, making it difficult to cope with the dynamic changes and sudden faults in complex operating environments. In addition, the maintenance cost of thermal power units is relatively high, and the equipment shutdown for maintenance has a great impact on power supply. How to maximize the system's adaptive ability and fault tolerance has become a key factor in optimizing the operation efficiency of thermal power units.

[0003] The existing signal adaptive control systems for thermal power units have poor global optimization ability, unable to perform dynamic adjustment and autonomous decision-making, reducing the stability and reliability of the units during long-term operation; in addition, the existing signal adaptive control systems for thermal power units cannot maintain good adaptability in uncertain and complex environments, reducing the full-life cycle efficiency of the units, and cannot automatically adjust the power generation capacity of the units or adjust the joint dispatching of multiple units. Summary of the Invention

[0004] The purpose of the present invention is to propose an intelligent signal adaptive control method and system for thermal power units to solve the defects existing in the prior art.

[0005] To achieve the above purpose, the present invention adopts the following technical solutions: An intelligent signal adaptive control method for thermal power units, comprising: Collect and preprocess the operation data obtained by various sensors of the thermal power unit, and extract feature information from the preprocessed operation data; Analyze each group of feature information and detect the fault information of the thermal power unit in real time; Establish a mathematical model of the thermal power unit for dynamic simulation, and synchronously monitor the detected fault information and operation data in real time; Predict the future state of the thermal power unit based on the operation data; Generate a control strategy according to the operation data, fault information, and future state prediction results to optimize the operation scheduling and resource allocation of the unit; Evaluate and adjust the control strategy, and adaptively adjust the strategy in real time; Perform fault tolerance processing based on the adaptive adjustment strategy when a fault occurs in the thermal power unit; Display the operation data and system status in real time, and feedback optimization suggestions according to requirements.

[0006] A further improvement of the present invention lies in that the specific steps of real-time detecting the fault information of the thermal power unit are as follows: S1.1: Extract the historical operation data of the thermal power unit from the external data center, perform data cleaning and normalization processing on each group of historical operation data, then use data analysis methods to select the characteristic data of each group related to the unit status and faults, and then extract the periodicity and frequency characteristic data of each group of historical operation data. Integrate the extracted characteristic data of each group to construct an operation characteristic set, and divide it into a training set, a test set, and a validation set; S1.2: Divide the training set into multiple small batches of training groups, construct a fault diagnosis model according to the LSTM model architecture and parameters. This model includes an input layer, an LSTM layer, a fully connected layer, and an output layer. Then, input the training groups into the fault diagnosis model in batches. Through the forward propagation of the fault diagnosis model, each training group of data is processed by the input gate, forget gate, and output gate in the LSTM layer in turn, and the final diagnosis data is generated; S1.3: After the LSTM layer inputs the generated diagnosis data into the fully connected layer for non-linear processing, the final diagnosis result is output through the output layer. Then, calculate the error value between the diagnosis result and the corresponding actual result through the mean absolute error function, and perform backpropagation of the calculated error value starting from the output layer of the fault diagnosis model to calculate the gradient value of the error value for each layer of the model. According to the gradient values of each layer, optimize the parameters of each layer of the fault diagnosis model through the gradient descent algorithm; S1.4: After each round of training is completed, input the validation set into the fault diagnosis model, and evaluate the average performance of the model through cross-validation. If the average performance of the model does not reach the preset threshold, adjust the hyperparameters of the fault diagnosis model and retrain and validate again. Otherwise, stop training, and use the test set data to detect the performance of the trained fault diagnosis model on unknown data. If the detection result reaches the preset expected value, input the new characteristic information into the fault diagnosis model, and perform forward propagation on each group of characteristic information through the model, and at the same time output the fault prediction labels and fault categories of each thermal power unit, where the label 0 represents a fault and the label 1 represents normal.

[0007] A further improvement of the present invention lies in that the specific steps of establishing a mathematical model of the thermal power unit for dynamic simulation are as follows: S2.1: Collect the design data and physical data of each thermal power unit, divide each thermal power unit into parts such as boilers, steam turbines, generators, and cooling systems, and then establish corresponding boiler dynamic models, steam turbine models, generator models, and cooling system models according to the physical laws of each part of each thermal power unit. Integrate each group of models into the corresponding thermal power unit model through data coupling; S2.2: Select a set of steady-state operating points of the state of the thermal power unit model under normal operating conditions, linearize the model based on the selected steady-state operating points, then simplify the linearized thermal power unit model, and then solve the dynamic equations of the thermal power unit model to obtain the system changes of each thermal power unit at different stages; S2.3: Analyze the system changes of each thermal power unit at different stages through graphical representation, evaluate the dynamic response of the system, and display the dynamic response of the thermal power unit under different operating conditions and control strategies in real time.

[0008] A further improvement of the present invention is that the specific steps for the generated control strategy to optimize the operation scheduling and resource allocation of the unit are as follows: S3.1: According to the real-time operation data, fault information, and future state prediction results of the thermal power unit, collect the energy dynamic changes of each thermal power unit and the interactions between each thermal power unit to generate a corresponding virtual ecological model. Use the control scheme of the current thermal power unit as the initial collaborative working strategy, and at the same time, with the goal of maximizing energy utilization efficiency, minimizing power loss, and extending the service life of equipment, construct a corresponding fitness function; P3.2: Initialize a group of populations according to the number of current thermal power units, and simulate the energy dynamic changes and interactions of each thermal power unit under the current collaborative working strategy based on the virtual ecological model. Then calculate the fitness values of each thermal power unit through the fitness function, and screen out the thermal power units below the preset selection threshold; P3.3: Copy the remaining thermal power units in the population, and according to the energy states of the copied thermal power units, adjust the collaborative working strategies of the copied groups of thermal power units by adjusting the operating parameters of the thermal power units. Then combine the adjusted thermal power units with the remaining storage units to form a new population, and use the newly generated population as the population for the next iteration; P3.4: Repeatedly update the population until the collaborative working strategies of each thermal power unit no longer change and the fitness values converge to the preset optimization goal. Then the virtual ecological model outputs the final collaborative working strategy, the energy states of each thermal power unit, and the energy interaction situation, and deploy this collaborative working strategy in the actual control of the thermal power unit, and at the same time perform real-time control.

[0009] A further improvement of the present invention is that the specific steps for evaluating and adjusting the control strategy and adaptively adjusting the strategy in real time are as follows: S4.1: Monitor the operating status of the thermal power unit in real time, extract the operating status of the thermal power unit in various states and the operations that can be taken in each state, construct a state set and an action set based on the collected state information and operation information, and then calculate the transfer probability of transferring to the next random state after collecting any action in the action set in each state of the state set; P4.2: Set the initial value of all state-action pairs to 0, collect the values of all state-action pairs, and construct the corresponding Q table. Then, at each moment, select an action to execute according to the current state and the information in the Q table; Set the initial value of all state-action pairs to 0, collect the values of all state-action pairs, and construct the corresponding Q table. Then, at each moment, select an action to execute according to the current state and the information in the Q table; value, and construct the corresponding Q table. Then, at each moment, select an action to execute according to the current state and the information in the Q table; P4.3: When selecting an action, generate a random number. If the random number is higher than the preset selection threshold, randomly select an action from the action set according to the calculated transfer probability; otherwise, select the action with the maximum value. After the action selection is completed, simulate each thermal power unit to enter the next state after executing the action; P4.4: Calculate the immediate reward in the corresponding state after executing any action according to the energy efficiency, power loss, and equipment life, and update the current state-action pair using the Q value update formula based on the obtained immediate reward and the generated next state, and update the Q table. Then, repeat the action selection and Q table update until the value change of the state-action pair in the Q table converges to the preset range; P4.5: After the iteration ends, traverse the values of each state-action pair in the Q table, select the state-action pair with the highest value as the optimal response strategy, and preset the operation strategy of the thermal power unit based on the feedback after each simulated state transition and the value update.

[0010] A further improvement of the present invention lies in that the specific steps for fault tolerance processing when a thermal power unit fails are as follows: S5.1: Receive the fault information. If there is a fault in the thermal power unit, identify the location of the fault source by comparing the difference between the state predicted by the thermal power unit model and the actual state; S5.2: When the location of the fault source is determined, analyze the fault component information at the current fault source location through the real-time acquisition data of the sensors at this location, isolate the detected fault components, switch to the standby components for operation, and adjust while maintaining the original control strategy.

[0011] The intelligent signal adaptive control system for thermal power units includes an acquisition and processing module, a monitoring and diagnosis module, a simulation module, a state prediction module, a control optimization module, a decision evaluation module, a fault tolerance module, and an interactive visualization module; The acquisition and processing module is used to acquire and preprocess the operation data obtained by various sensors of the thermal power unit, and extract feature information from the preprocessed operation data; The monitoring and diagnosis module is used to analyze each group of feature information and detect the fault information of the thermal power unit in real time; The simulation and modeling module is used to establish a mathematical model of the thermal power unit for dynamic simulation, and synchronize the detected fault information and operation data in real time; The state prediction module is used to predict the future state of the thermal power unit based on the operation data; The control and optimization module is used to generate a control strategy based on the operation data, fault information and future state prediction results, and optimize the operation scheduling and resource allocation of the unit; The decision-making and evaluation module is used to evaluate and adjust the control strategy generated by the control and optimization module, and adaptively adjust the strategy in real time; The fault tolerance module is used to perform fault tolerance processing when the thermal power unit fails; The interactive visualization module is used to provide real-time display of operation data and system status, and feedback optimization suggestions according to requirements.

[0012] A further improvement of the present invention is that the specific steps for the monitoring and diagnosis module to detect the fault information of the thermal power unit in real time are as follows: S1.1: Extract the historical operation data of the thermal power unit from the external data center, perform data cleaning and normalization processing on each group of historical operation data, then use data analysis methods to select the feature data of each group related to the unit status and faults, and then extract the periodicity and frequency of each group of historical operation data. Each group of feature data is integrated to construct an operation feature set, and it is divided into a training set, a test set and a validation set; S1.2: Divide the training set into multiple small batches of training groups, construct a fault diagnosis model according to the LSTM model architecture and parameters. This model includes an input layer, an LSTM layer, a fully connected layer and an output layer. Then, input the training groups into the fault diagnosis model batch by batch. Through the forward propagation of the fault diagnosis model, each training group of data is processed by the input gate, forget gate and output gate in the LSTM layer in turn, and the final diagnosis data is generated; S1.3: After the LSTM layer inputs the generated diagnosis data into the fully connected layer for non-linear processing, the final diagnosis result is output through the output layer. Then, calculate the error value between the diagnosis result and the corresponding actual result through the mean absolute error function, and perform backpropagation of the calculated error value starting from the output layer of the fault diagnosis model to calculate the gradient value of the error value for each layer of the model. According to the gradient values of each layer, optimize the parameters of each layer of the fault diagnosis model through the gradient descent algorithm; S1.4: After each round of training is completed, the validation set is input into the fault diagnosis model, and the average performance of the model is evaluated through the cross-validation method. If the average performance of the model does not reach the preset threshold, the hyperparameters of the fault diagnosis model are adjusted and training and validation are carried out again. Otherwise, the training is stopped, and the performance of the trained fault diagnosis model on unknown data is detected using the test set data. If the detection result reaches the preset expected value, the new feature information is input into the fault diagnosis model, and forward propagation is performed on each group of feature information through the model, and at the same time, the fault prediction labels and fault categories of each thermal power unit are output, where the label 0 represents a fault and the label 1 represents normal.

[0013] A further improvement of the present invention is that the specific steps for the simulation module to establish a mathematical model of the thermal power unit for dynamic simulation are as follows: S2.1: The simulation module collects the design data and physical data of each thermal power unit, divides each thermal power unit into parts such as boilers, steam turbines, generators, and cooling systems, and then establishes corresponding boiler dynamic models, steam turbine models, generator models, and cooling system models according to the physical laws of each part of each thermal power unit, and integrates each group of models into the corresponding thermal power unit model through data coupling; S2.2: Select a set of steady-state operating points of the state of the thermal power unit model under normal operating conditions, perform linearization processing on the model based on the selected steady-state operating points, then simplify the linearized thermal power unit model, and then solve the dynamic equations of the thermal power unit model to obtain the system changes of each thermal power unit at different stages; S2.3: Analyze the system changes of each thermal power unit at different stages through graphing, evaluate the dynamic response of the system, and display the dynamic response of the thermal power unit under different working conditions and control strategies in real time.

[0014] A further improvement of the present invention is that the specific steps for the control optimization module to optimize the operation scheduling and resource allocation of the unit are as follows: S3.1: According to the real-time operation data, fault information, and future state prediction results of the thermal power unit, collect the energy dynamic changes of each thermal power unit and the interactions between each thermal power unit to generate a corresponding virtual ecological model. Take the current control scheme of the thermal power unit as the initial collaborative working strategy, and at the same time, with the goal of maximizing energy utilization efficiency, minimizing power loss, and extending the service life of equipment, construct a corresponding fitness function; P3.2: Initialize a group of populations according to the current number of thermal power units, simulate the energy dynamic changes and interactions of each thermal power unit under the current collaborative working strategy based on the virtual ecological model, then calculate the fitness values of each thermal power unit through the fitness function, and then screen out the thermal power units below the preset selection threshold; P3.3: Duplicate the remaining thermal power units in the population, and according to the energy states of the duplicated thermal power units, adjust the collaborative working strategies of each group of duplicated thermal power units by adjusting the operating parameters of the thermal power units. Then, combine the adjusted thermal power units with the remaining storage units to form a new population, and use the newly generated population as the population for the next iteration; P3.4: Repeatedly update the population until the collaborative working strategies of each thermal power unit no longer change and the fitness value converges to the preset optimization goal. Then, the virtual ecological model outputs the final collaborative working strategy, the energy states of each thermal power unit, and the energy interaction situation, and deploys this collaborative working strategy in the actual control of thermal power units while performing real-time control.

[0015] Compared with the prior art, the present invention has at least the following beneficial technical effects: The present invention collects the energy dynamic changes of each thermal power unit and the interactions between each thermal power unit to generate a corresponding virtual ecological model, uses the current control scheme of the thermal power unit as the initial collaborative working strategy, and constructs a corresponding fitness function with the goals of maximizing energy utilization efficiency, minimizing power loss, and extending the service life of the equipment. Initialize a group of populations according to the number of current thermal power units, simulate the energy dynamic changes and interactions of each thermal power unit under the current collaborative working strategy based on the virtual ecological model, calculate the fitness values of each thermal power unit through the fitness function, then screen out the thermal power units below the preset selection threshold, duplicate the remaining thermal power units in the population, and according to the energy states of the duplicated thermal power units, adjust the collaborative working strategies of each group of duplicated thermal power units by adjusting the operating parameters of the thermal power units. Then, combine the adjusted thermal power units with the remaining storage units to form a new population, and use the newly generated population as the population for the next iteration. Repeatedly update the population until the collaborative working strategies of each thermal power unit no longer change and the fitness value converges to the preset optimization goal. Then, the virtual ecological model outputs the final collaborative working strategy, the energy states of each thermal power unit, and the energy interaction situation, and deploys this collaborative working strategy in the actual control of thermal power units while performing real-time control, which can improve the global optimization ability of the system, enhance the fault tolerance and recovery ability of the unit, achieve dynamic adjustment and autonomous decision-making, and improve the stability and reliability of the unit during long-term operation.

[0016] The present invention constructs a state set and an action set based on the collected state information and operation information. Then, it calculates the transition probability of transferring to the next random state after taking any action in the action set under each state in the state set. The initial Q-value of all state-action pairs is set to 0. It collects the Q-values of all state-action pairs and constructs a corresponding Q-table. Then, at each moment, according to the current state and the information in the Q-table, it selects an action to execute. When selecting an action, a random number is generated. If the random number is higher than the preset selection threshold, a random action is selected from the action set according to the calculated transition probability; otherwise, the action with the largest Q-value is selected. After the action selection is completed, it simulates that each thermal power unit enters the next state after executing this action, calculates the immediate reward in the corresponding state after executing any action based on the energy efficiency, power loss, and equipment life, and updates the current state-action pair using the Q-value update formula according to the obtained immediate reward and the generated next state, and updates the Q-table. Then, it repeats the action selection and Q-table update until the change in the Q-value of the state-action pair in the Q-table converges to the preset range. After the iteration ends, it traverses the Q-values of each state-action pair in the Q-table and selects the state-action pair with the highest value as the optimal response strategy. At the same time, based on the feedback after each simulated state transition and the Q-value update, it pre-sets the operation strategy of the thermal power unit, which can improve the adaptability of the system in an uncertain and complex environment, help improve the full-life cycle efficiency of the unit, reduce the long-term operation cost, ensure the continuous and efficient operation of the system, and at the same time can automatically adjust the power generation capacity of the unit or adjust the coordinated dispatch of multiple units to avoid energy waste or equipment damage. Description of the Drawings

[0017] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1 It is a structural block diagram of the intelligent signal adaptive control system for thermal power units proposed by the present invention. Detailed Embodiments

[0019] In the following text, only some exemplary embodiments are simply described. As those skilled in the art can recognize, the described embodiments can be modified in various different ways without departing from the spirit or scope of the present invention. Therefore, the drawings and the description are considered to be exemplary in nature rather than restrictive.

[0020] It should be understood that when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not preclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their combinations.

[0021] It should also be understood that the terms used in the specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0022] It should be further understood that the term "and / or" used in the specification of the present invention and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0023] Schematic diagrams of various structures according to the disclosed embodiments of the present invention are shown in the drawings. These figures are not drawn to scale, where certain details are enlarged for the purpose of clear expression and certain details may be omitted. The shapes of various regions and layers shown in the figures, as well as their relative sizes and positional relationships, are merely exemplary and may deviate in practice due to manufacturing tolerances or technical limitations. Those skilled in the art can additionally design regions / layers with different shapes, sizes and relative positions according to actual requirements.

[0024] Embodiments of the present invention will be described in detail below with reference to the drawings.

[0025] Embodiment 1 Refer to Figure 1 , an intelligent thermal power unit signal adaptive control system, including an acquisition and processing module, a monitoring and diagnosis module, a simulation and simulation module, a state prediction module, a control optimization module, a decision-making and evaluation module, a fault tolerance module, and an interactive visualization module.

[0026] The acquisition and processing module is used to acquire and preprocess the operation data obtained by various sensors of the thermal power unit, and extract feature information from the preprocessed operation data.

[0027] The monitoring and diagnosis module is used to analyze each group of feature information and detect the fault information of the thermal power unit in real time.

[0028] Specifically, the monitoring and diagnosis module extracts the historical operation data of thermal power units from an external data center, cleans and normalizes the historical operation data of each group. Then, it selects the characteristic data of each group related to the unit status and faults using data analysis methods, and extracts the periodicity and frequency characteristic data of each group of historical operation data. The extracted characteristic data of each group are integrated to construct an operation characteristic set, which is divided into a training set, a test set, and a validation set. The training set is divided into multiple small batches of training groups. A fault diagnosis model is constructed according to the LSTM model architecture and parameters. The model includes an input layer, an LSTM layer, a fully connected layer, and an output layer. Then, the training groups are input into the fault diagnosis model in batches. Through the forward propagation of the fault diagnosis model, the data of each training group are sequentially processed by the input gate, forget gate, and output gate in the LSTM layer, and the final diagnosis data are generated. The LSTM layer inputs the generated diagnosis data into the fully connected layer for non-linear processing, and then outputs the final diagnosis result through the output layer. Then, the error value between the diagnosis result and the corresponding actual result is calculated through the mean absolute error function, and the calculated error value is backpropagated from the output layer of the fault diagnosis model to calculate the gradient value of the error value for each layer of the model. Based on the gradient values of each layer, the parameters of each layer of the fault diagnosis model are optimized through the gradient descent algorithm. After each round of training is completed, the validation set is input into the fault diagnosis model, and the average performance of the model is evaluated through the cross-validation method. If the average performance of the model does not reach the preset threshold, the hyperparameters of the fault diagnosis model are adjusted and the training and validation are carried out again. Otherwise, the training is stopped, and the performance of the trained fault diagnosis model on unknown data is detected using the test set data. If the detection result reaches the preset expected value, new characteristic information is input into the fault diagnosis model, and forward propagation is performed on the characteristic information of each group through the model, and at the same time, the fault prediction labels and fault categories of each thermal power unit are output, where the label 0 represents a fault and the label 1 represents normal.

[0029] The simulation module is used to establish a mathematical model of the thermal power unit for dynamic simulation, and to synchronize the monitored fault information and operation data in real time.

[0030] Specifically, the simulation module collects the design data and physical data of each thermal power unit, divides each thermal power unit into parts such as boilers, steam turbines, generators, and cooling systems, and then establishes corresponding boiler dynamic models, steam turbine models, generator models, and cooling system models according to the physical laws of each part of each thermal power unit. The models of each group are integrated into the corresponding thermal power unit model through data coupling. A set of steady-state operating points of the state of the thermal power unit model under normal operating conditions is selected, and the model is linearized based on the selected steady-state operating points. Then, the linearized thermal power unit model is simplified. After that, the dynamic equations of the thermal power unit model are solved to obtain the system changes of each thermal power unit at different stages. The system changes of each thermal power unit at different stages are analyzed through graphical representation, the dynamic response of the system is evaluated, and the dynamic response of the thermal power unit under different working conditions and control strategies is displayed in real time.

[0031] Embodiment 2 Refer to Figure 1 , the intelligent adaptive control system for thermal power unit signals includes an acquisition and processing module, a monitoring and diagnosis module, a simulation module, a state prediction module, a control optimization module, a decision-making and evaluation module, a fault tolerance module, and an interactive visualization module.

[0032] The state prediction module is used to predict the future state of the thermal power unit based on the operation data.

[0033] The control optimization module is used to generate control strategies to optimize the operation scheduling and resource allocation of the unit according to the operation data, fault information, and future state prediction results.

[0034] Specifically, according to the real-time operation data, fault information, and future state prediction results of thermal power units, collect the energy dynamic changes of each thermal power unit and the interactions between each thermal power unit to generate a corresponding virtual ecological model. Use the current control scheme of the thermal power unit as the initial collaborative working strategy. At the same time, with the goal of maximizing energy utilization efficiency, minimizing power loss, and extending the service life of equipment, construct a corresponding fitness function. Initialize a group of populations based on the number of current thermal power units, and simulate the energy dynamic changes and interactions of each thermal power unit under the current collaborative working strategy based on the virtual ecological model. Then calculate the fitness values of each thermal power unit through the fitness function, screen out the thermal power units below the preset selection threshold, copy the remaining thermal power units in the population, and adjust the collaborative working strategies of the copied groups of thermal power units by adjusting the operating parameters of the thermal power units according to the energy states of the copied thermal power units. After that, combine the adjusted thermal power units with the remaining storage units to form a new population, and use the newly generated population as the population for the next iteration. Repeatedly update the population until the collaborative working strategies of each thermal power unit no longer change and the fitness values converge to the preset optimization goal. Then the virtual ecological model outputs the final collaborative working strategy, the energy states of each thermal power unit, and the energy interaction situation, and deploy this collaborative working strategy in the actual control of thermal power units while performing real-time control.

[0035] The decision-making evaluation module is used to evaluate and adjust the control strategy generated by the control optimization module and adaptively adjust the strategy in real time.

[0036] Specifically, monitor the operating state of the thermal power unit in real time, extract the states of the thermal power unit in various states and the operations that can be taken in each state, and construct a state set and an action set based on the collected state information and operation information. Then calculate the transition probability of transferring to the next random state after taking any action in the action set under each state in the state set, and set the initial value of all state-action pairs to 0. Collect all state-action pairs, and construct a corresponding Q table. Then at each moment, according to the current state and the information in the Q table, select an action to execute. When selecting an action, generate a random number. If the random number is higher than the preset selection threshold, randomly select an action from the action set according to the calculated transition probability. Otherwise, select the action with the largest value. After the action selection is completed, simulate each thermal power unit to enter the next state after executing this action, calculate the immediate reward in the corresponding state after executing any action according to the energy efficiency, power loss, and equipment life, and update the current state-action pair and the Q table using the Q value update formula based on the obtained immediate reward and the generated next state. Then repeat the action selection and Q table update until the initial value of the state-action pairs in the Q table is 0, collect all the values of the state-action pairs, and construct a corresponding Q table. Then at each moment, according to the current state and the information in the Q table, select an action to execute. When selecting an action, generate a random number. If the random number is higher than the preset selection threshold, randomly select an action from the action set according to the calculated transition probability. Otherwise, select the action with the largest value. After the action selection is completed, simulate each thermal power unit to enter the next state after executing this action, calculate the immediate reward in the corresponding state after executing any action according to the energy efficiency, power loss, and equipment life, and update the current state-action pair and the Q table using the Q value update formula based on the obtained immediate reward and the generated next state. Then repeat the action selection and Q table update until the values of the state-action pairs in the Q table no longer change. The value change converges to the preset range. After the iteration is completed, the state-action pairs in the Q table are traversed. value, and select the state-action pair with the highest value as the optimal response strategy, and at the same time, according to the feedback and The value is updated and the operation strategy of the thermal power unit is pre-set.

[0037] The fault tolerance module is used to perform fault tolerance processing when a fault occurs in a thermal power unit.

[0038] Specifically, the fault tolerance module receives the fault information output by the monitoring and diagnosis module. If there is a fault in the thermal power unit, the location of the fault source is identified by comparing the difference between the state predicted by the thermal power unit model and the actual state. After the location of the fault source is determined, the real-time data collected by the sensor at that location is used to analyze the fault component information at the current fault source location, and the detected faulty components are isolated and switched to spare components for operation, while the original control strategy is maintained for adjustment.

[0039] The interactive visualization module is used to provide real-time operation data and system status display, and provide optimization suggestions based on demand feedback.

[0040] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the attached claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any figure mark in the claims should not be regarded as limiting the claims involved.

[0041] In addition, it should be understood that although this specification is described in accordance with the implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation modes that can be understood by those skilled in the art. The above content is only to illustrate the technical idea of ​​the present invention, and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution according to the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. An intelligent thermal power unit signal adaptive control method, characterized in that Including: Collect and preprocess the operation data obtained from various sensors of the thermal power unit, and extract feature information from the preprocessed operation data; Analyze each group of feature information and detect the fault information of the thermal power unit in real time; Establish a mathematical model of the thermal power unit for dynamic simulation, and synchronize the detected fault information and operation data in real time; Predict the future state of the thermal power unit based on the operation data; Generate a control strategy based on the operation data, fault information, and future state prediction results, and optimize the operation scheduling and resource allocation of the unit; Evaluate and adjust the control strategy, and adaptively adjust the strategy in real time; Perform fault tolerance processing based on the adaptive adjustment strategy when the thermal power unit fails; Display the operation data and system status in real time, and feedback optimization suggestions according to requirements.

2. The intelligent thermal power unit signal adaptive control method according to claim 1, characterized in that The specific steps for detecting the fault information of the thermal power unit in real time are as follows: S1.1: Extract the historical operation data of the thermal power unit from the external data center, perform data cleaning and normalization processing on each group of historical operation data, then use data analysis methods to select the feature data of each group related to the unit status and faults, and then extract the periodicity and frequency of each group of historical operation data. Integrate the extracted feature data of each group to construct an operation feature set, and divide it into a training set, a test set, and a validation set; S1.2: Divide the training set into multiple small batches of training groups, construct a fault diagnosis model according to the LSTM model architecture and parameters. This model includes an input layer, an LSTM layer, a fully connected layer, and an output layer. Then, input the training groups into the fault diagnosis model in batches. Through the forward propagation of the fault diagnosis model, each training group of data is processed by the input gate, forget gate, and output gate in the LSTM layer in turn, and the final diagnosis data is generated; S1.3: After the LSTM layer inputs the generated diagnosis data into the fully connected layer for nonlinear processing, the final diagnosis result is output through the output layer. Then, calculate the error value between the diagnosis result and the corresponding actual result through the mean absolute error function, and perform backpropagation of the calculated error value starting from the output layer of the fault diagnosis model to calculate the gradient value of the error value for each layer of the model. According to the gradient values of each layer, optimize the parameters of each layer of the fault diagnosis model through the gradient descent algorithm; S1.4: After each round of training is completed, input the validation set into the fault diagnosis model, and evaluate the average performance of the model through the cross-validation method. If the average performance of the model does not reach the preset threshold, adjust the hyperparameters of the fault diagnosis model and retrain and validate again. Otherwise, stop training, and use the test set data to detect the performance of the trained fault diagnosis model on unknown data. If the detection result reaches the preset expected value, input the new feature information into the fault diagnosis model, and perform forward propagation on each group of feature information through the model, and output the fault prediction labels and fault categories of each thermal power unit at the same time. Among them, the label 0 represents a fault, and the label 1 represents normal.

3. The intelligent thermal power unit signal adaptive control method according to claim 2, wherein The specific steps for establishing a mathematical model of the thermal power unit for dynamic simulation are as follows: S2.1: Collect the design data and physical data of each thermal power unit, divide each thermal power unit into parts such as boilers, steam turbines, generators, and cooling systems, then establish corresponding boiler dynamic models, steam turbine models, generator models, and cooling system models according to the physical laws of each part of each thermal power unit, and integrate each group of models into the corresponding thermal power unit model through data coupling; S2.2: Select a set of steady-state operating points of the state of the thermal power unit model under normal operating conditions, linearize the model based on the selected steady-state operating points, then simplify the linearized thermal power unit model, and then solve the dynamic equations of the thermal power unit model to obtain the system changes of each thermal power unit at different stages; S2.3: Analyze the system changes of each thermal power unit at different stages through graphical representation, evaluate the dynamic response of the system, and display the dynamic response of the thermal power unit under different operating conditions and control strategies in real time.

4. The intelligent thermal power unit signal adaptive control method according to claim 3, characterized in that The specific steps for the generated control strategy to optimize the operation scheduling and resource allocation of the unit are as follows: S3.1: According to the real-time operation data, fault information, and future state prediction results of the thermal power unit, collect the energy dynamic changes of each thermal power unit and the interactions between each thermal power unit to generate a corresponding virtual ecological model. Use the current control scheme of the thermal power unit as the initial collaborative working strategy, and at the same time, with the goal of maximizing energy utilization efficiency, minimizing power loss, and extending the service life of equipment, construct a corresponding fitness function; P3.2: Initialize a group of populations according to the current number of thermal power units, simulate the energy dynamic changes and interactions of each thermal power unit under the current collaborative working strategy based on the virtual ecological model, then calculate the fitness values of each thermal power unit through the fitness function, and then screen out the thermal power units below the preset selection threshold; P3.3: Copy the remaining thermal power units in the population, and according to the energy states of the copied thermal power units, adjust the collaborative working strategies of the copied groups of thermal power units by adjusting the operating parameters of the thermal power units. Then combine the adjusted thermal power units with the remaining storage units to form a new population, and use the newly generated population as the population for the next iteration; P3.4: Repeatedly update the population until the collaborative working strategies of each thermal power unit no longer change and the fitness values converge to the preset optimization goal. Then the virtual ecological model outputs the final collaborative working strategy, the energy states of each thermal power unit, and the energy interaction situation, and deploy this collaborative working strategy in the actual control of the thermal power unit, and at the same time perform real-time control.

5. The intelligent thermal power unit signal adaptive control method according to claim 4, wherein The specific steps for evaluating and adjusting the control strategy and adaptively adjusting the strategy in real time are as follows: S4.1: Monitor the operating state of the thermal power unit in real time, extract the states of the thermal power unit in each state and the operations that can be taken in each state, and construct a state set and an action set based on the collected state information and operation information. Then calculate the transition probability of transferring to the next random state after taking any action in the action set for each state in the state set; P4.2: Set the initial value of for all state-action pairs to 0, collect the values of all state-action pairs, and construct the corresponding Q-table. Then, at each moment, select an action to execute according to the current state and the information in the Q-table; P4.3: When selecting an action, generate a random number. If the random number is higher than the preset selection threshold, randomly select an action from the action set according to the calculated transition probability; otherwise, select the action with the largest value. After the action selection is completed, simulate each thermal power unit to execute the action and enter the next state; P4.4: Calculate the immediate reward in the corresponding state after performing any action based on energy efficiency, power loss, and device lifespan, and update the current state-action pair using the Q-value update formula according to the obtained immediate reward and the generated next state, and update the Q-table. Then repeat action selection and Q-table update until the change in the value of the state-action pair in the Q-table converges within a preset range; P4.5: After the iteration ends, traverse the values of each state-action pair in the Q-table, and select the state-action pair with the highest value as the optimal response strategy. At the same time, based on the feedback after each simulated state transition and the value update, preset the operation strategy of the thermal power unit. And update the value, and preset the operation strategy of the thermal power unit. value update, preset the operation strategy of the thermal power unit.

6. The intelligent thermal power unit signal adaptive control method according to claim 1, characterized in that, The specific steps for fault tolerance processing when a thermal power unit fails are as follows: S5.1: Receive the fault information. If there is a fault in the thermal power unit, identify the location of the fault source by comparing the difference between the predicted state and the actual state of the thermal power unit model. S5.2: After determining the location of the fault source, analyze the fault component information at the current fault source location through the real-time acquisition data of the sensors at this location, isolate the detected faulty components, switch to the standby components for operation, and adjust while maintaining the original control strategy.

7. Intelligent adaptive control system for signals of thermal power generating units, characterized in that, It includes an acquisition and processing module, a monitoring and diagnosis module, a simulation module, a state prediction module, a control optimization module, a decision-making and evaluation module, a fault tolerance module, and an interactive visualization module. The acquisition and processing module is used to acquire and preprocess the operation data obtained by various sensors of the thermal power unit, and extract feature information from the preprocessed operation data. The monitoring and diagnosis module is used to analyze each group of feature information and detect the fault information of the thermal power unit in real time. The simulation module is used to establish a mathematical model of the thermal power unit for dynamic simulation, and synchronize the detected fault information and operation data in real time. The state prediction module is used to predict the future state of the thermal power unit based on the operation data. The control optimization module is used to generate a control strategy based on the operation data, fault information, and future state prediction results, and optimize the operation scheduling and resource allocation of the unit. The decision-making and evaluation module is used to evaluate and adjust the control strategy generated by the control optimization module, and adaptively adjust the strategy in real time. The fault tolerance module is used to perform fault tolerance processing when a fault occurs in the thermal power unit. The interactive visualization module is used to provide real-time display of operation data and system status, and feedback optimization suggestions according to requirements.

8. The intelligent thermal power unit signal adaptive control system according to claim 7, wherein, The specific steps for the monitoring and diagnosis module to detect the fault information of the thermal power unit in real time are as follows: S1.1: Extract the historical operation data of the thermal power unit from the external data center, perform data cleaning and normalization processing on each group of historical operation data, then use data analysis methods to select each group of feature data related to the unit state and faults, and then extract the periodicity and frequency of each group of historical operation data. Each group of feature data is integrated to construct an operation feature set, and it is divided into a training set, a test set, and a validation set. S1.2: Divide the training set into multiple small batches of training groups. Construct a fault diagnosis model according to the LSTM model architecture and parameters. This model includes an input layer, an LSTM layer, a fully connected layer, and an output layer. Then, input the training groups into the fault diagnosis model in batches. Through the forward propagation of the fault diagnosis model, each group of training data is processed by the input gate, forget gate, and output gate in the LSTM layer in turn, and the final diagnosis data is generated. S1.3: The LSTM layer inputs the generated diagnosis data into the fully connected layer for nonlinear processing, and then outputs the final diagnosis result through the output layer. Then, calculate the error value between the diagnosis result and the corresponding actual result through the mean absolute error function, and perform backpropagation of the calculated error value starting from the output layer of the fault diagnosis model to calculate the gradient value of the error value for each layer of the model. According to the gradient values of each layer, optimize the parameters of each layer of the fault diagnosis model through the gradient descent algorithm. S1.4: After each round of training is completed, input the validation set into the fault diagnosis model, and evaluate the average performance of the model through the cross-validation method. If the average performance of the model does not reach the preset threshold, adjust the hyperparameters of the fault diagnosis model and retrain and validate it. Otherwise, stop the training, and use the test set data to detect the performance of the trained fault diagnosis model on unknown data. If the detection result reaches the preset expected value, input the new feature information into the fault diagnosis model, and perform forward propagation on each group of feature information through the model, and at the same time output the fault prediction labels and fault categories of each thermal power unit, where the label 0 represents a fault and the label 1 represents normal.

9. The intelligent thermal power unit signal adaptive control system according to claim 8, characterized in that, The specific steps for the simulation module to establish a mathematical model of the thermal power unit for dynamic simulation are as follows: S2.1: The simulation module collects the design data and physical data of each thermal power unit, divides each thermal power unit into parts such as boilers, steam turbines, generators, and cooling systems, and then establishes corresponding boiler dynamic models, steam turbine models, generator models, and cooling system models according to the physical laws of each part of each thermal power unit, and integrates each group of models into the corresponding thermal power unit model through data coupling; S2.2: Select a set of steady-state operating points of the thermal power unit model under normal operating conditions, linearize the model based on the selected steady-state operating points, then simplify the linearized thermal power unit model, and then solve the dynamic equations of the thermal power unit model to obtain the system changes of each thermal power unit at different stages; S2.3: Analyze the system changes of each thermal power unit at different stages through graphical methods, evaluate the dynamic response of the system, and display the dynamic response of the thermal power unit under different operating conditions and control strategies in real time.

10. The intelligent thermal power unit signal adaptive control system according to claim 9, characterized in that The specific steps for the control optimization module to optimize the operation scheduling and resource allocation of the unit are as follows: S3.1: According to the real-time operation data, fault information, and future state prediction results of the thermal power unit, collect the energy dynamic changes of each thermal power unit and the interactions between each thermal power unit to generate the corresponding virtual ecological model. Take the control scheme of the current thermal power unit as the initial collaborative working strategy, and at the same time, with the goal of maximizing energy utilization efficiency, minimizing power loss, and extending the service life of equipment, construct the corresponding fitness function; P3.2: Initialize a group of populations according to the number of current thermal power units, and simulate the energy dynamic changes and interactions of each thermal power unit under the current collaborative working strategy based on the virtual ecological model. Then calculate the fitness values of each thermal power unit through the fitness function, and screen out the thermal power units below the preset selection threshold; P3.3: Copy the remaining thermal power units in the population, and adjust the collaborative working strategy of each copied thermal power unit by adjusting the operating parameters of the thermal power unit according to the energy state of the copied thermal power unit. Then combine the adjusted thermal power units with the remaining storage units to form a new population, and use the newly generated population as the population for the next iteration; P3.4: Repeatedly perform population update until the collaborative working strategies of each thermal power unit no longer change and the fitness value converges to the preset optimization goal. After that, the virtual ecological model outputs the final collaborative working strategy, the energy states of each thermal power unit, and the energy interaction situation, and deploys this collaborative working strategy in the actual control of thermal power units while performing real-time control.

Images