Online simulation optimization system of thermal power plant and maintenance system of thermal power plant
Through the factory-wide situational awareness, process model and model self-evolution module of the online simulation optimization system, the simulation model of the thermal power plant is monitored and updated in real time, which solves the problem of inaccurate simulation results in the existing technology, and improves the autonomy of the simulation system and Accuracy improves the operating status of thermal power plants.
Patent Information
- Application Number
- CN202410995369.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-24
- Publication Date
- 2024-11-29
AI Technical Summary
The existing thermal power plant simulation systems cannot accurately reflect the dynamic changes of the system, lack adaptability, and cannot update the simulation model in real time, resulting in inaccurate simulation results and difficult to cope with external interference and equipment aging problems.
An online simulation optimization system for thermal power plants was designed, including the factory-wide situational awareness module, process model module and model self-evolution module. Through monitoring, prediction and decision-making units, the deviation simulation model is selected. Use particle swarm optimization algorithms and machine learning algorithms to update the simulation model to improve the accuracy of simulation results.
The simulation input data is consistent with the actual operating state, and a strategic solution is provided to improve the operating state of the thermal power plant, improve the accuracy and autonomy of the simulation results, and reduce manual intervention and operation costs.
Smart Images

Figure CN119048049A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of thermal power plant simulators, and in particular to an online simulation optimization system of a thermal power plant and a maintenance system of a thermal power plant. Background Art
[0002] In order to comply with increasingly stringent environmental regulations, thermal power plants need to continuously optimize their operating modes, reduce emissions, and improve clean production levels. Today’s thermal power plants have complex system structures involving multiple subsystems and equipment. Traditional manual operations and empirical decision-making are difficult to cope with the complexity and constant dynamic changes of the system. The rapid development of computer technology and automation technology provides new methods for simulation and optimization of power plants.
[0003] Existing simulation systems often face the following major defects when applied in complex systems such as thermal power plants: First, the simulation model is built based on static models or historical data (non-real-time data), so it cannot accurately reflect the dynamic changes of the system under different operating conditions; second, the simulation model is usually run offline and cannot obtain data and update it during real-time operation, so the simulation results are inaccurate; third, it lacks adaptive capabilities and cannot automatically adjust the simulation model parameters according to real-time data, so the simulation model is difficult to cope with external interference; fourth, the simulation model is difficult to maintain due to aging equipment or changes in operating conditions in thermal power plants. Summary of the invention
[0004] The technical problem to be solved by the present invention is to overcome the defect of poor accuracy of simulation results of traditional thermal power plant simulation systems in the prior art, and to provide an online simulation optimization system for a thermal power plant and a maintenance system for a thermal power plant.
[0005] The present invention solves the above technical problems through the following technical solutions:
[0006] In a first aspect, the present invention provides an online simulation optimization system for a thermal power plant, comprising: an online simulation system and an online optimization system, wherein the online simulation system comprises a plant-wide situation awareness module, and the online optimization system comprises a process model module and a model self-evolution module;
[0007] The whole plant situation awareness module is used to couple the real operating data of the thermal power plant with the online simulation system to obtain the prediction analysis results of the operation status in the future period, and provide a strategy for the operation status of the thermal power plant according to the real operating data and the prediction analysis results; the online simulation system synchronously inputs the input data of each real system of the corresponding thermal power plant;
[0008] The process model module is used to screen out several target simulation models that cause deviations from the online simulation system when the difference information between the actual operating condition data and the output operating condition data of the online simulation system exceeds a set threshold;
[0009] The model self-evolution module is used to trigger an online optimization mechanism for the target simulation model to calculate a target empirical coefficient, and update the online simulation system according to the target empirical coefficient.
[0010] Preferably, the whole plant situation awareness module includes a monitoring unit, a prediction unit and a decision-making unit.
[0011] The monitoring unit is used to monitor the thermal power plant in real time through the real operating condition data and the output operating condition data of the simulation model;
[0012] The prediction unit is used to predict the operating state of the thermal power plant in a preset period in the future through a machine learning algorithm based on historical operating condition data and current operating condition data, and generate the prediction analysis result; the historical operating condition data is obtained by coupling the real operating condition data of the historical period with the online simulation system, and the current operating condition data is obtained by coupling the real operating condition data of the current period with the online simulation system;
[0013] The decision-making unit is used to provide a strategy for the operating status of the thermal power plant according to the prediction analysis results and the actual operating condition data.
[0014] Preferably, the process model module includes a difference monitoring unit and a model selection unit;
[0015] The difference monitoring unit is used to monitor whether the difference information between the output working condition data and the real working condition data exceeds the set threshold, and when it exceeds the set threshold, the model selection unit is called, and if it does not exceed the set threshold, the output working condition data continues to be monitored;
[0016] The model selection unit is used to separate the physical model causing the difference and the corresponding empirical coefficient system from the online simulation system as the target simulation model for online optimization.
[0017] Preferably, the model self-evolution module includes an algorithm optimization unit and a model evaluation unit;
[0018] The algorithm optimization unit is used to optimize the empirical coefficients of the target simulation model using a particle swarm optimization algorithm to obtain the target empirical coefficients;
[0019] The model evaluation unit is used to apply the target empirical coefficient to the online simulation system to update the parameters of the target simulation model.
[0020] Preferably, the online simulation system further includes a fault diagnosis module;
[0021] The fault diagnosis module is used to perform preprocessing, feature extraction and fault diagnosis on the real working condition data of the current period and the original data obtained by coupling the online simulation system to obtain a diagnosis report result.
[0022] Preferably, the fault diagnosis module includes a preprocessing unit, a feature extraction unit and a fault diagnosis unit;
[0023] The preprocessing unit is used to perform a preset process on the raw data to obtain initial data, wherein the preset process includes at least one of a cleaning process, a filtering process, a conversion process and a normalization process;
[0024] The feature extraction unit is used to perform feature extraction and feature selection on the initial data according to the analysis method FT-PCA and the recursive feature elimination method SVM-RFE to obtain target data;
[0025] The fault diagnosis unit is used to identify the fault type of the target data using a machine learning algorithm.
[0026] Preferably, the prediction unit includes a storage subunit, a model training subunit and a prediction subunit;
[0027] The storage subunit is used to store the pre-processed output operating condition data of the online simulation system in a pre-established data warehouse;
[0028] The model training subunit is used to train the neural network model using a machine learning algorithm to obtain a prediction model;
[0029] The prediction subunit is used to predict the operating status of the thermal power plant in a future preset time period using the prediction model.
[0030] Preferably, the decision unit includes a strategy calling subunit and a strategy generating subunit;
[0031] The strategy calling subunit is used to call the strategy plan from a pre-built knowledge base including the operating procedures, emergency plans and optimization strategies of various equipment in the thermal power plant according to the prediction analysis results and the real working condition data;
[0032] The strategy generation subunit is used to analyze the prediction analysis results and the actual operating condition data using a simulated annealing algorithm SA and a genetic algorithm GA to generate a strategy solution.
[0033] Preferably, the actual operating condition data includes at least one of boiler parameter data, turbine parameter data, generator parameter data, auxiliary equipment parameter data and environmental parameter data.
[0034] In a second aspect, the present invention provides a maintenance system for a thermal power plant, the maintenance system comprising a thermal power plant operation system, an intelligent industrial base, and an online simulation optimization system for the thermal power plant as described in the first aspect;
[0035] The intelligent industrial base is communicatively connected with the thermal power plant operation system and the online simulation optimization system;
[0036] The intelligent industrial base is used to obtain the real working condition data of the thermal power plant operation system and send the real working condition data to the online simulation optimization system;
[0037] The online simulation optimization system is used to generate prediction and analysis results based on the actual operating condition data and the output operating condition data of the simulation model, and to provide a strategy plan for the operating status of the thermal power plant based on the actual operating condition data and the prediction and analysis results, and to feed back the strategy plan to the thermal power plant operation system through the intelligent industrial base.
[0038] The positive progressive effects of the present invention are: providing an online simulation optimization system for a thermal power plant and a maintenance system for the thermal power plant, using the plant-wide situation awareness module in the online simulation system to ensure that the simulation input data and the actual operating status data of the thermal power plant are consistent, ensuring the validity of the simulation results, so as to provide a strategy plan for the operating status of the thermal power plant based on the real operating data and the prediction analysis results, and improve the operating status of the thermal power plant; using the process model module and the model self-evolution module in the online optimization system to continuously optimize and update the physical model with deviations, so as to improve the accuracy of the simulation results. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 This is a module diagram of an online simulation optimization system for a thermal power plant according to Example 1 of the present invention.
[0040] Figure 2 This is a module diagram of a maintenance system for a thermal power plant according to Embodiment 2 of the present invention.
[0041] Figure 3 This is a schematic diagram of the application of the maintenance system of a thermal power plant according to Example 2 of the present invention. DETAILED DESCRIPTION
[0042] The present invention is further described below by way of examples, but the present invention is not limited to the scope of the examples.
[0043] Example 1
[0044] The online simulation optimization system of the thermal power plant in this embodiment is as follows: Figure 1As shown, the online simulation optimization system 10 includes: an online simulation system 100 and an online optimization system 200, the online simulation system 100 includes a plant situation awareness module 110, and the online optimization system 200 includes a process model module 210 and a model self-evolution module 220;
[0045] The whole plant situation awareness module 110 is used to couple the real operating data of the thermal power plant with the online simulation system to obtain the prediction analysis results of the operation status in the future period, and provide a strategy plan for the operation status of the thermal power plant according to the real operating data and the prediction analysis results; the online simulation system synchronously inputs the input data of each real system of the corresponding thermal power plant;
[0046] The process model module 210 is used to select a plurality of target simulation models that cause the deviation from the online simulation system when the difference information between the actual working condition data and the output working condition data of the online simulation system exceeds a set threshold;
[0047] The model self-evolution module 220 is used to trigger an online optimization mechanism for the target simulation model to calculate the target empirical coefficient, and update the online simulation system according to the target empirical coefficient.
[0048] The actual operating condition data includes at least one of boiler parameter data, turbine parameter data, generator parameter data, auxiliary equipment parameter data and environmental parameter data.
[0049] Specifically, the online simulation system 100 obtains the real topological parameters of the thermal power plant through the virtual DCS system, establishes an initial online simulation system according to the real topological parameters, assigns values to the online simulation system 100 based on the real operating state data of the thermal power plant, obtains the initial operating state of the online simulation system 100 of the thermal power plant, and then performs real-time calculations on the online simulation system 100 based on the steady-state operating conditions of the thermal power plant and the preset constraint boundary conditions, and generates several simulation models that operate synchronously with the thermal power plant. The whole plant situation awareness module 110 predicts the operation of the thermal power plant in the future preset period based on the historical operating data and the current operating data obtained after coupling the real operating data of the thermal power plant with the online simulation system 100, obtains the prediction analysis results, and provides strategic solutions for the thermal power plant through the prediction analysis results and the real operating data. In this embodiment, the output operating data of the online simulation system 100 can make up for the real operating data that cannot be collected by the sensors of the real thermal power plant, and the two sources of data complement each other.
[0050] The main function of the online optimization system 200 is to continuously optimize, improve the performance of the physical simulation model of the online simulation system 100, improve the model simulation accuracy, reduce manual intervention, and improve the system autonomy and intelligence. When the process parameter error of the online simulation system 100 and the real thermal power plant synchronous operation exceeds the set threshold, the process model module 210 traces back to the upstream online simulation system 100 according to the data flow, and automatically extracts the target simulation model that causes the deviation in the system. The model self-evolution module 220 performs back calculation optimization on the empirical coefficients of the target simulation model. After the back calculation optimization is completed, the empirical coefficients in the target simulation model are updated to the online simulation system 100, thereby improving the simulation accuracy of the online simulation system 100 and the accuracy of the simulation results.
[0051] In this embodiment, the plant situation awareness module proposes operating parameter adjustment suggestions based on the prediction results to prevent potential problems and improve operating efficiency; through online optimization technology, the physical model of the online simulation system can be optimized quickly and accurately, and the synchronization and consistency between the online simulation system and the actual thermal power plant operating status can be enhanced; based on the online optimization system, a large amount of debugging time and cost can be saved, and the system design and debugging process can be accelerated.
[0052] The plant situation awareness module 110 includes a monitoring unit 111, a prediction unit 112 and a decision-making unit 113.
[0053] A monitoring unit 111 is used to monitor the thermal power plant in real time through real operating condition data and output operating condition data of the simulation model;
[0054] The prediction unit 112 is used to predict the operating state of the thermal power plant in a preset period in the future based on the historical operating condition data and the current operating condition data by a machine learning algorithm, and generate a prediction analysis result; the historical operating condition data is obtained by coupling the real operating condition data of the historical period with the online simulation system, and the current operating condition data is obtained by coupling the real operating condition data of the current period with the online simulation system;
[0055] The decision-making unit 113 is used to provide a strategy for the operation status of the thermal power plant according to the prediction analysis results and the actual operating condition data.
[0056] Specifically, the monitoring unit 111 receives the real working condition data of the thermal power plant in real time through various sensors and data acquisition equipment, and receives the output working condition data of the simulation model in the online simulation system 100 running synchronously, and uses the real working condition data and the output working condition data to monitor the thermal power plant in real time. The real working condition data includes boiler parameter data, turbine parameter data, generator parameter data, auxiliary equipment parameter data and environmental parameter data. The boiler parameter data includes power generation, speed, torque, bearing temperature, etc., and the environmental parameter data includes ambient temperature, ambient humidity and meteorological conditions, etc.
[0057] The prediction unit 112 cleans and normalizes the historical operating condition data and the current operating condition data to ensure data quality, uses a machine learning algorithm regression analysis combined with a neural network (RNN) to train a prediction analysis model based on the historical operating condition data, and then uses a cross-validation method to ensure the accuracy of the prediction analysis results of the prediction analysis model.
[0058] Exemplarily, the following is the training process of the predictive analysis model: Step 1. Define the predictive analysis model structure; the number of nodes in the input layer is equal to the number of features, each hidden layer includes a certain number of nodes, and an activation function such as ReLU (f(x) = max(0, x)) is used. The output layer has only one node, no activation function is used or a linear activation function is used. Step 2. Define the loss function and optimizer; for regression tasks, the commonly used loss function is the mean square error (MSE), which is defined as: Among them, y^ is the model prediction value, y is the true value, N is the number of samples, and the optimizer uses stochastic gradient descent (SGD) and introduces normal cloud distribution. The normal cloud optimization process includes: 1) initialization parameters, 2) selection of learning rate, 3) selection of noise standard deviation, and 4) iterative calculation. Step 3. Training the predictive analysis model; randomly initialize the weights and biases of the network, input data into the network, calculate the predicted value y^ through each layer, calculate the gradient of the loss function with respect to the weights and biases, use the chain rule, and use the optimizer to update the weights according to the calculated gradient. Step 4. Cross-validation; in order to ensure the generalization ability of the predictive analysis model, k cross-validation is used, and the predictive analysis model is determined by the average results of k validations. Step 5. Hyperparameter tuning; adjust hyperparameters such as learning rate, number of hidden layers, number of nodes, and regularization parameters to obtain the best predictive analysis model. Step 6. Predictive analysis model evaluation; use the performance of the independent test set model that did not participate in the training to ensure that the model has good generalization ability.
[0059] The decision unit 113 selects a strategy solution from a knowledge base containing a large amount of rules and expert knowledge, or generates a strategy solution by calculation using an optimization algorithm, based on the prediction analysis results and the actual operating condition data.
[0060] The process model module 210 includes a difference monitoring unit 211 and a model selection unit 212;
[0061] The difference monitoring unit 211 is used to monitor whether the difference information between the output working condition data and the actual working condition data exceeds a set threshold. When the difference information exceeds the set threshold, the model selection unit 212 is called. If the difference information does not exceed the set threshold, the output working condition data continues to be monitored.
[0062] The model selection unit 212 is used to separate the physical model causing the difference and the corresponding empirical coefficient system from the online simulation system 100 as the target simulation model for online optimization.
[0063] Specifically, the difference monitoring unit 211 monitors the difference information between the output operating condition data of the online simulation system 100 during continuous operation and the actual operating condition data of the thermal power plant. When the difference information exceeds the set threshold, the online optimization mechanism is triggered, and the model selection unit 212 automatically identifies and locks the physical model that causes the difference, and separates it from the online simulation system 100 as the target of online optimization.
[0064] The model self-evolution module 220 includes an algorithm optimization unit 221 and a model evaluation unit 222;
[0065] The algorithm optimization unit 221 is used to optimize the empirical coefficients of the target simulation model by using a particle swarm optimization algorithm to obtain target empirical coefficients;
[0066] The model evaluation unit 222 is used to apply the target empirical coefficient to the online simulation system to update the parameters of the target simulation model.
[0067] Specifically, the algorithm optimization unit 221 uses the support vector product optimized particle swarm algorithm (PSO-SVM) to adjust the empirical coefficient in the target simulation model so as to minimize the prediction error and obtain the target empirical coefficient. Exemplarily, the following is a detailed process for optimizing the empirical coefficient: Step 1. Establishing an SVM model; using a portion of the collected real operating data as a training set, establishing an SVM model to predict the output index data of a thermal power plant. Step 2. Particle swarm optimization; using the PSO algorithm to optimize the empirical coefficient in the online simulation system 100 to minimize the gap between the SVM predicted value and the actual value, and repeatedly updating the particle position and velocity through the steps of initializing the particle swarm, evaluating the particles, updating the particle position and velocity, and iterative optimization, until the preset number of iterations is reached or the fitness value meets the stop condition. Step 3. Evaluate and update the SVM model; apply the optimized empirical coefficient to the output predicted data in the online simulation system 100 for comparison with the real data, and evaluate the degree of optimization of the empirical coefficient of the target simulation model. Step 4. Iterate and monitor the target simulation model; repeat the above steps 1-3 periodically or when the performance of the target simulation model decreases to continuously optimize the target simulation model, and adjust the parameters of the target simulation model in a timely manner according to the changes in the operating conditions of the thermal power plant to ensure the accuracy and timeliness of the target simulation model. The model evaluation unit 222 updates the target simulation model according to the target empirical coefficient and optimizes the online simulation system 100.
[0068] The online simulation system 100 further includes a fault diagnosis module 120;
[0069] The fault diagnosis module 120 is used to perform preprocessing, feature extraction and fault diagnosis on the real working condition data of the current period and the raw data obtained by coupling the online simulation system to obtain a diagnosis report result. In this embodiment, the operation status of the thermal power plant is monitored and diagnosed in real time through intelligent fault diagnosis, and potential problems are detected and identified in time; predictive maintenance is performed on the thermal power plant system or equipment to facilitate maintenance personnel to take preventive measures in time and extend the service life of the thermal power plant equipment.
[0070] Specifically, the fault diagnosis module 120 performs analysis based on the output of the diagnosis model in combination with the rule base and the expert system, generates a detailed diagnosis report result, and proposes specific maintenance or repair measures.
[0071] The fault diagnosis module 120 includes a preprocessing unit 121, a feature extraction unit 122 and a fault diagnosis unit 123;
[0072] A preprocessing unit 121, configured to perform a preset process on the raw data to obtain initial data, wherein the preset process includes at least one of a cleaning process, a filtering process, a conversion process, and a normalization process;
[0073] A feature extraction unit 122 is used to extract and select features from the initial data according to the analysis method FT-PCA and recursive feature elimination SVM-RFE to obtain target data;
[0074] The fault diagnosis unit 123 is used to identify the fault type of the target data using a machine learning algorithm.
[0075] Specifically, the preprocessing unit 121 cleans, filters and converts the raw data, removes noise and fills in missing data. The feature extraction unit 122 uses the principal component analysis FT-PCA after Fourier transform optimization to extract features from the preprocessed data, and uses support vector product optimized recursive feature elimination (SVM-RFE) to perform feature selection to reduce data dimensions, highlight key features and select the most useful features for fault diagnosis. Exemplarily, the specific process of the feature extraction FT-PCA part is as follows: Step 1. Preprocess the data, including standardization (mean is 0, variance is 1) and missing value processing (filling or deleting missing values). Step 2. Apply Fourier transform to each sample data to convert the data from the time domain to the frequency domain. Step 3. Extract the main frequency features from the results of the Fourier transform. Step 4. Combine the extracted frequency features with the original data or other features to form a new feature set. Step 5. Standardize the new feature set to ensure that the mean of each feature is 0 and the variance is 1; Step 6. Apply PCA to the standardized feature set to reduce the dimension and extract the main components; Step 7. Analyze the results of PCA, check the explained variance of the main components, and determine the number of main components; Step 8. Adjust the parameters of Fourier transform and PCA according to the results to continuously optimize the model. Through the above steps 1-8, the frequency characteristics of the data are effectively extracted by Fourier transform, and PCA is combined for dimension reduction and feature extraction, thereby optimizing the effect of PCA. The specific process of the feature selection SVM-RFE part is as follows: Step 1. Initialize the SVM model, use all available features to train the SVM model, and determine the decision boundary, which is determined by the weight vector w and the bias term b. Step 2. Calculate the feature weight w. Step 3. Sort the features; Sort the features according to the absolute value of the weight vector w. Step 4. Eliminate features; Remove one or more features at the end after sorting. Step 5. Repeat the training and elimination process, retrain the SVM model using the remaining features, calculate the weight vector again, rank the features, and eliminate the least important features.
[0076] The prediction unit 112 includes a storage subunit, a model training subunit and a prediction subunit;
[0077] A storage subunit, used for storing the pre-processed output working condition data of the online simulation system in a pre-established data warehouse;
[0078] A model training subunit is used to train a neural network model using a machine learning algorithm to obtain a prediction model;
[0079] The prediction subunit is used to use the prediction model to predict the operating status of the thermal power plant in a preset time period in the future.
[0080] The decision unit 113 includes a policy calling subunit and a policy generating subunit;
[0081] The strategy calling subunit is used to call the strategy plan from the pre-built knowledge base including the operating procedures, emergency plans and optimization strategies of various equipment in the thermal power plant according to the prediction analysis results and the actual working condition data;
[0082] The strategy generation subunit is used to analyze the prediction analysis results and actual operating condition data using the simulated annealing algorithm SA and the genetic algorithm GA to generate a strategy solution.
[0083] This embodiment provides an online simulation optimization system for a thermal power plant, which uses the plant situation awareness module in the online simulation system to ensure that the simulation input data and the actual operating status data of the thermal power plant are consistent, and to ensure the validity of the simulation results, so as to provide a strategy plan for the operating status of the thermal power plant based on the real operating data and the prediction analysis results, and improve the operating status of the thermal power plant; the process model module and the model self-evolution module in the online optimization system are used to continuously optimize and update the physical model with deviations, so as to improve the accuracy of the simulation results.
[0084] Example 2
[0085] This embodiment provides a maintenance system for a thermal power plant, such as Figure 2 and Figure 3 As shown, the maintenance system includes a thermal power plant operation system 400, an intelligent industrial base 300, and an online simulation optimization system 10 of a thermal power plant as in Example 1;
[0086] The intelligent industrial base 300 is connected in communication with the thermal power plant operation system 400 and the online simulation optimization system;
[0087] The intelligent industrial base 300 is used to obtain the real working condition data of the thermal power plant operation system 400 and send the real working condition data to the online optimization simulation system 10;
[0088] The online simulation optimization system 10 is used to generate prediction analysis results based on the actual operating condition data and the output operating condition data of the simulation model, and to provide a strategy plan for the operating status of the thermal power plant based on the actual operating condition data and the prediction analysis results, and to feed back the strategy plan to the thermal power plant operation system 400 through the intelligent industrial base 300.
[0089] This embodiment provides a maintenance system for a thermal power plant, which uses a real-time communication interface to connect the online simulation optimization system and the intelligent industrial base of the thermal power plant to ensure the real-time transmission of data and the accuracy of data interaction; uses the online simulation system to realize dynamic simulation of the thermal power plant to ensure that the simulation result is the same as the actual operating state, so as to improve the reliability and credibility of the simulation; uses the online optimization system to respond to the changes in input and output in a timely manner, and updates the mechanism model with deviations in time; optimizes the simulation model, improves resource utilization, and reduces operating costs.
[0090] Although the specific embodiments of the present invention are described above, it should be understood by those skilled in the art that this is only for illustration and the protection scope of the present invention is defined by the appended claims. Those skilled in the art may make various changes or modifications to these embodiments without departing from the principles and essence of the present invention, but these changes and modifications all fall within the protection scope of the present invention.
Claims
1. An online simulation optimization system for thermal power plants, It is characterized in that include: An online simulation system and an online optimization system, wherein the online simulation system includes a plant-wide situation awareness module, and the online optimization system includes a process model module and a model self-evolution module; The whole plant situation awareness module is used to couple the real operating data of the thermal power plant with the online simulation system to obtain the prediction analysis results of the operating status in the future period, and provide a strategy plan for the operating status of the thermal power plant according to the real operating data and the prediction analysis results; The online simulation system synchronously inputs the input data of each real system of the corresponding thermal power plant; The process model module is used to screen out several target simulation models that cause deviations from the online simulation system when the difference information between the actual operating condition data and the output operating condition data of the online simulation system exceeds a set threshold; The model self-evolution module is used to trigger an online optimization mechanism for the target simulation model to calculate a target empirical coefficient, and update the online simulation system according to the target empirical coefficient.
2. The online simulation optimization system for a thermal power plant according to claim 1, It is characterized in that The whole plant situation awareness module includes a monitoring unit, a prediction unit and a decision-making unit. The monitoring unit is used to monitor the thermal power plant in real time through the real operating condition data and the output operating condition data of the simulation model; The prediction unit is used to predict the operating state of the thermal power plant in a future preset period of time based on the historical operating condition data and the current operating condition data through a machine learning algorithm to generate the prediction analysis result; The historical operating condition data is obtained by coupling the real operating condition data of the historical period with the online simulation system, and the current operating condition data is obtained by coupling the real operating condition data of the current period with the online simulation system; The decision-making unit is used to provide a strategy for the operating status of the thermal power plant according to the prediction analysis results and the actual operating condition data.
3. The online simulation optimization system for a thermal power plant according to claim 1, It is characterized in that The process model module includes a difference monitoring unit and a model selection unit; The difference monitoring unit is used to monitor whether the difference information between the output working condition data and the real working condition data exceeds the set threshold, and when it exceeds the set threshold, the model selection unit is called, and if it does not exceed the set threshold, the output working condition data continues to be monitored; The model selection unit is used to separate the physical model causing the difference and the corresponding empirical coefficient system from the online simulation system as the target simulation model for online optimization.
4. The online simulation optimization system for a thermal power plant according to claim 1, It is characterized in that The model self-evolution module includes an algorithm optimization unit and a model evaluation unit; The algorithm optimization unit is used to optimize the empirical coefficients of the target simulation model using a particle swarm optimization algorithm to obtain the target empirical coefficients; The model evaluation unit is used to apply the target empirical coefficient to the online simulation system to update the parameters of the target simulation model.
5. The online simulation optimization system for a thermal power plant according to claim 1, It is characterized in that The online simulation system also includes a fault diagnosis module; The fault diagnosis module is used to perform preprocessing, feature extraction and fault diagnosis on the real working condition data of the current period and the original data obtained by coupling the online simulation system to obtain a diagnosis report result.
6. The online simulation optimization system for a thermal power plant according to claim 5, It is characterized in that The fault diagnosis module includes a preprocessing unit, a feature extraction unit and a fault diagnosis unit; The preprocessing unit is used to perform a preset process on the raw data to obtain initial data, wherein the preset process includes at least one of a cleaning process, a filtering process, a conversion process and a normalization process; The feature extraction unit is used to perform feature extraction and feature selection on the initial data according to the analysis method FT-PCA and the recursive feature elimination method SVM-RFE to obtain target data; The fault diagnosis unit is used to identify the fault type of the target data using a machine learning algorithm.
7. The online simulation optimization system for a thermal power plant according to claim 2, It is characterized in that The prediction unit includes a storage subunit, a model training subunit and a prediction subunit; The storage subunit is used to store the pre-processed output operating condition data of the online simulation system in a pre-established data warehouse; The model training subunit is used to train the neural network model using a machine learning algorithm to obtain a prediction model; The prediction subunit is used to predict the operating status of the thermal power plant in a future preset time period using the prediction model.
8. The online simulation optimization system for a thermal power plant according to claim 2, It is characterized in that The decision-making unit includes a strategy calling subunit and a strategy generating subunit; The strategy calling subunit is used to call the strategy plan from a pre-built knowledge base including the operating procedures, emergency plans and optimization strategies of various equipment in the thermal power plant according to the prediction analysis results and the real working condition data; The strategy generation subunit is used to analyze the prediction analysis results and the actual operating condition data using a simulated annealing algorithm SA and a genetic algorithm GA to generate a strategy solution.
9. The online simulation optimization system for a thermal power plant according to claim 1, It is characterized in that The actual operating condition data includes at least one of boiler parameter data, turbine parameter data, generator parameter data, auxiliary equipment parameter data and environmental parameter data.
10. A maintenance system for a thermal power plant, It is characterized in that The maintenance system comprises a thermal power plant operation system, an intelligent industrial base and an online simulation optimization system for a thermal power plant according to any one of claims 1 to 9; The intelligent industrial base is communicatively connected with the thermal power plant operation system and the online simulation optimization system; The intelligent industrial base is used to obtain the real working condition data of the thermal power plant operation system and send the real working condition data to the online simulation optimization system; The online simulation optimization system is used to generate prediction and analysis results based on the actual operating condition data and the output operating condition data of the simulation model, and to provide a strategy plan for the operating status of the thermal power plant based on the actual operating condition data and the prediction and analysis results, and to feed back the strategy plan to the thermal power plant operation system through the intelligent industrial base.
