Intelligent Optimization Control Method for Supercritical Thermal Power Generation Units

The intelligent optimization control method for supercritical thermal power generation units addresses degraded performance by adapting control strategies based on real-time and historical data, improving efficiency and reliability through optimized system operation and reduced emissions.

JP7953990B2Active Publication Date: 2026-11-19SHANGAN POWER PLANT OF HUANENG INT POWER CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
JP2025248080
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2025-01-06
Filing Date
2025-12-12
Publication Date
2026-11-19
Estimated Expiration
2045-12-12

AI Technical Summary

Technical Problem

Supercritical thermal power generation units face degraded control performance due to aging equipment and changing characteristics, leading to increased operator intervention, coal consumption, ammonia slip, urea consumption, and reduced AGC and primary frequency control performance, with existing control methods failing to adapt to these changes.

Method used

An intelligent optimization control method that acquires real-time and historical operating parameters, trains a model, performs optimization analysis, and adjusts control policies based on feature extraction and control strategies to optimize system operation.

Benefits of technology

Improves the efficiency and reliability of thermal power generation units by reducing emissions, optimizing system operation, and achieving energy conservation and emission reduction targets, while minimizing operator intervention and enhancing equipment performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007953990000001
    Figure 0007953990000001
Patent Text Reader

Abstract

This invention provides an intelligent optimization control method for supercritical thermal power generation units. [Solution] The solution includes acquiring real-time operating parameters and historical operating parameters of each power generation unit in the power plant, outputting initial data collectively, selecting and acquiring a first model from a model database with reference to a selection command, training and optimizing the first model using the initial data, and outputting a predetermined analysis model, performing an optimization analysis on the initial data based on the predetermined analysis model and with reference to predetermined evaluation indicators, outputting the optimization analysis results, extracting features from the optimization analysis results, selecting and acquiring control policies and control methods that map to features extracted from a policy database with reference to a predetermined feature policy comparison table, and adjusting the operating parameters of each power generation unit based on the control policies and control methods.
Need to check novelty before this filing date? Find Prior Art

Description

TECHNICAL FIELD

[0001] The present invention relates to the technical field of thermal power generation, in particular to an intelligent optimal control method for a supercritical thermal power generation unit. BACKGROUND ART

[0002] Supercritical thermal power generation units have complex production processes. With the continuous long-term operation of the supercritical thermal power generation unit, the equipment gradually ages, and its characteristics change significantly. As a result, the performance of control systems for important parameters such as main steam temperature, steam pressure, and environmental protection related parameters degrades, which leads to frequent manual intervention by operators, and greatly increases the operation load and monitoring burden. Furthermore, in many cases, the intervention of operators on the control system is carried out from the perspective of safe operation of the supercritical thermal power generation unit, and the economy of the unit is neglected. For this reason, units that have been operating for a long time universally have the problem of maintaining the set values of important systems such as main steam temperature, steam pressure, and denitration at a low level to prevent the reference values of parameters from being exceeded. This phenomenon leads to an increase in coal consumption of the unit, an increase in ammonia slip and urea consumption, and further results in the degradation of AGC (Automatic Generation Control) and primary frequency control performance of the unit, which may cause a reduction in the economy of the unit.

[0003] At present, many control methods or systems for supercritical power generation units are constructed by standard modules such as PID, and their parameters cannot be self-adjusted according to changes in unit characteristics. For this reason, after the unit has been operated for a long time, the problem of degraded control performance arises.

[0004] Accordingly, the present invention provides an intelligent optimal control method for a supercritical thermal power generation unit. SUMMARY OF THE INVENTION PROBLEMS TO BE SOLVED BY THE INVENTION

[0005] This invention provides an intelligent optimization control method for supercritical thermal power generation units. This method ensures the safe operation of the unit and equipment, reduces operator intervention in critical systems, lowers the monitoring burden, unlocks the unit's potential, and improves economic efficiency. [Means for solving the problem]

[0006] The present invention provides an intelligent optimization control method for a supercritical thermal power generation unit. The intelligent optimization control method for a supercritical thermal power generation unit includes steps 1 to 4. Step 1 involves acquiring real-time operating parameters and historical operating parameters for each power generation unit in the power plant and outputting the initial data collectively. Step 2 involves selecting and acquiring a first model from a model database with reference to a selection command, training and optimizing the first model using the initial data, and outputting a predetermined analysis model. Step 3 involves performing an optimization analysis on the initial data based on the predetermined analysis model and with reference to predetermined evaluation indicators, and outputting the optimization analysis results. Step 4 involves extracting features from the optimization analysis results, selecting and acquiring control policies and control methods that map to features extracted from a policy database with reference to a predetermined feature policy comparison table, and adjusting the operating parameters of each power generation unit based on the control policies and control methods.

[0007] Preferably, acquiring real-time operating parameters for each power generation unit within a power plant includes acquiring user monitoring request information, selecting and acquiring appropriate predetermined monitoring equipment from an equipment database based on the monitoring request information, and acquiring operating parameter data for each power generation unit in real time based on the predetermined monitoring equipment, and outputting the real-time operating parameter data with reference to a timestamp corresponding to the operating parameter data.

[0008] Preferably, step 1 further includes extracting features from the real-time operation parameter data, constructing and obtaining a first feature set, mapping and obtaining corresponding historical operation parameters from a historical database based on the first feature set, and outputting historical operation parameter data.

[0009] Preferably, prior to outputting the initial data in a batch, the intelligent optimization control method for the supercritical thermal power generation unit includes selecting and obtaining a preprocessing method to map from a method database based on the monitoring request information and a first feature set, and performing preprocessing on the real-time operation parameter data and historical operation parameter data based on the preprocessing method.

[0010] Preferably, the above process of selecting and obtaining a first model from a model database with reference to a selection command includes obtaining a user-entered model manual selection command and a system-generated model automatic selection command, outputting a model selection command with reference to the priority corresponding to each command, and selecting and obtaining a first model to map from the model database based on the model selection command.

[0011] Preferably, training and optimizing a first model using the initial data includes performing data partitioning on the initial data using a predetermined data partitioning method to obtain a training dataset, a validation dataset, and a test dataset, and training and optimizing the first model based on the training dataset, the validation dataset, and the test dataset.

[0012] Preferably, step 3 includes: extracting features from the monitoring request information and constructing a second feature set based on the extracted features; obtaining corresponding indicator selection factors based on the second feature set and with reference to a predetermined feature-factor comparison table, and constructing and obtaining a selection factor set; selecting and obtaining a first indicator to map to each indicator selection factor in the selection factor set from the indicator database with reference to a predetermined factor-indicator mapping table, and constructing and obtaining an indicator candidate pool; obtaining an automatic indicator selection command generated by the system using a predetermined algorithm in real time, and simultaneously obtaining a human-inputted indicator selection command via a predetermined port in real time, and outputting an indicator selection command set with reference to priority information corresponding to each command; selecting and obtaining a predetermined evaluation indicator to map from the indicator candidate pool based on the indicator selection command set, and constructing and obtaining an evaluation indicator set; and the evaluation The method further includes inputting a set of valuation indicators into a predetermined analysis model, performing setting adjustments and updates to the predetermined analysis model with reference to a predetermined self-adaptive algorithm, simultaneously outputting model format request data after the setting update, simultaneously obtaining corresponding data format information for all data included in the initial data and outputting an initial data format comparison table, selecting and obtaining a predetermined format conversion method from a method database to map the model format request information to the initial data format comparison table, performing format conversion on the initial data based on the model format request data and the initial data format comparison table, and using the predetermined format conversion method, and outputting the data to be analyzed, and performing optimization analysis on the data to be analyzed through the predetermined analysis model after the setting update with reference to the set of evaluation indicators, and outputting the optimization analysis results.

[0013] Preferably, step 4 involves inputting predetermined operating threshold conditions into the predetermined analysis model to generate predetermined boundary conditions, acquiring self-generated optimization indicators output by the predetermined analysis model based on the predetermined boundary conditions and artificial optimization indicators obtained by artificial selection, outputting an optimization indicator set based on the self-generated optimization indicators and the artificial optimization indicators, inputting the initial data into the predetermined analysis model to perform predictive analysis and acquiring a first prediction result by a variable structure predictive control method and a second prediction result by a generalized predictive control method, creating a process based on the optimization indicator set, the first prediction result and the second prediction result, and referring to predetermined policies and methods, creating and acquiring an intelligent control policy and an intelligent control method, and mapping relationship between the optimization indicator set, the intelligent control policy and the intelligent control method. This includes constructing and obtaining an indicator-policy-method comparison table based on the relationship; performing feature extraction on the optimization analysis results, constructing and obtaining an optimization feature set based on the extracted features; obtaining optimization indicators to map to each feature in the optimization feature set by referring to a predetermined feature-indicator mapping table, and outputting an optimization indicator set awaiting execution; obtaining intelligent control policies and intelligent control methods to map to each optimization indicator in the optimization indicator set awaiting execution by referring to the indicator-policy-method comparison table, and outputting an execution policy-method list; and performing control and adjustment on the corresponding operating parameters in each power generation unit based on the intelligent control policies and intelligent control methods in the execution policy-method list, and generating process data in real time and feeding it back to a predetermined data analysis model. [Effects of the Invention]

[0014] This invention provides an intelligent optimization control method for supercritical thermal power generation units. The invention achieves intelligent optimization control of the power generation unit's operating parameters by training and optimizing a model, as well as by referencing feature extraction and control strategy selection, through the acquisition of real-time and historical operating parameters. The invention enables intelligent optimization analysis based on real-time and historical data to select the optimal control strategy and method. This improves the efficiency of the power plant's thermal power generation units, reduces emissions, optimizes system operation, and enables the achievement of energy conservation and emission reduction targets, thereby improving the economics and reliability of power plant operation. [Brief explanation of the drawing]

[0015] To more clearly explain the technical concepts of the present invention or the prior art, the drawings used in the examples or descriptions of the prior art are briefly introduced below. The drawings in the following description are some examples of the present invention, and those skilled in the art can obtain other drawings based on these without adding any creative work. [Figure 1] This is a flowchart of an intelligent optimization control method for a supercritical thermal power generation unit, provided by an embodiment of the present invention. [Modes for carrying out the invention]

[0016] To further clarify the purpose, technical proposal, and advantages of the present invention, the technical proposal of the present invention will be clearly and completely described below with reference to the drawings of the present invention. Note that the embodiments described are only a selection of embodiments of the present invention and do not include all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without additional creative work are all within the scope of the protection of the present invention.

[0017] As shown in Figure 1, the intelligent optimization control method for a supercritical thermal power generation unit provided by an embodiment of the present invention includes steps 1 to 4. Step 1 involves acquiring real-time operating parameters and historical operating parameters of each power generation unit in the power plant and outputting the initial data collectively. Step 2 involves selecting and acquiring a first model from a model database with reference to a selection command, training and optimizing the first model using the initial data, and outputting a predetermined analysis model. Step 3 involves performing an optimization analysis on the initial data based on the predetermined analysis model and with reference to a predetermined evaluation index, and outputting the optimization analysis results. Step 4 involves extracting features from the optimization analysis results, selecting and acquiring a control policy and control method to map to the extracted features from a policy database with reference to a predetermined feature policy comparison table, and adjusting the operating parameters of each power generation unit based on the control policy and control method.

[0018] In this embodiment, real-time operating parameters refer to the current operating parameters (temperature, pressure, output, etc.) of each power generation unit within the power plant.

[0019] In this embodiment, historical operating parameters refer to the operating parameter data for each power generation unit within the power plant over a past period, and are used for analysis and comparison (for example, the load curve and temperature changes of the power generation unit over the past week).

[0020] In this embodiment, initial data refers to data formed by integrating acquired real-time operating parameters and historical operating parameters, and which serves as the starting point for optimization analysis (for example, including real-time load, historical failure records, and power generation unit model information).

[0021] In this embodiment, a selection command refers to a command or rule for selecting an appropriate model from a model database.

[0022] In this embodiment, the model database refers to a database storing various models, and is used to select an appropriate model based on a selection command.

[0023] In this embodiment, the first model refers to an initial data model selected from the model database based on the selection command, and is used for training and optimization.

[0024] In this embodiment, the predetermined analysis model refers to a model that has undergone training and optimization, and is used for analysis and prediction of initial data (for example, it is a neural network model that has undergone data training and optimization).

[0025] In this embodiment, the predetermined evaluation index refers to an index for evaluating the optimization analysis result, and is helpful for judging the quality of system performance.

[0026] In this embodiment, the optimization analysis result refers to a result obtained by performing optimization using the predetermined analysis model and the evaluation index, and is used to determine a next control measure.

[0027] In this embodiment, the predetermined feature-strategy comparison table contains the correspondence between features and strategies, and is preset.

[0028] In this embodiment, the strategy database refers to a database storing various control strategies and methods.

[0029] In this embodiment, the control strategies and methods are control strategies and methods selected based on the feature extraction result, and are used to adjust the operating parameters of each power generation unit.

[0030] The implementation principle and beneficial effects of this embodiment are as follows. This invention realizes intelligent optimized control of power generation unit operating parameters by training and optimizing a model, as well as referencing feature extraction and control strategy selection, through the acquisition of real-time and historical operating parameters. This invention enables intelligent optimization analysis based on real-time and historical data to select the optimal control strategy and method. This improves the efficiency of thermal power generation units in power plants, reduces emissions, optimizes system operation, and enables the achievement of energy conservation and emission reduction targets, thereby improving the economics and reliability of power plant operation.

[0031] In the intelligent optimization control method for a supercritical thermal power generation unit provided in an embodiment of the present invention, acquiring real-time operating parameters of each power generation unit in the power plant includes acquiring user monitoring request information, selecting and acquiring appropriate predetermined monitoring equipment from an equipment database based on the monitoring request information, and acquiring operating parameter data for each power generation unit in real time based on the predetermined monitoring equipment, and outputting the real-time operating parameter data with reference to a timestamp corresponding to the operating parameter data.

[0032] In this embodiment, the monitoring request information refers to the request information such as parameters and frequency that the user needs from the monitoring equipment (for example, the user must acquire load, temperature, and vibration data of the power generation unit every minute).

[0033] In this embodiment, the equipment database refers to a database that stores various monitoring equipment information and is used to select appropriate monitoring equipment (for example, including technical parameters and model information of various sensors and monitoring devices).

[0034] In this embodiment, the predetermined monitoring equipment refers to appropriate monitoring equipment selected from the equipment database based on monitoring demand information (for example, selecting one temperature sensor and one vibration monitoring device to constitute the predetermined monitoring equipment).

[0035] In this embodiment, the operating parameter data refers to the real-time operating parameter data of each power generation unit (for example, the load of power generation unit A is 100 MW, the temperature is 300°C, etc.).

[0036] In this embodiment, a timestamp refers to information that indicates the time point of data collection and ensures the time-series and accuracy of the data (for example, each data point is assigned a timestamp indicating the data collection time).

[0037] In this embodiment, real-time operation parameter data refers to operation parameter data of each power generation unit acquired in real time based on predetermined monitoring equipment, and is output with reference to a timestamp (for example, load, temperature, and vibration data of power generation unit A acquired every minute, and the corresponding timestamp information).

[0038] The implementation principle and beneficial effects of this embodiment are as follows. The present invention enables real-time monitoring of the operating status of power generation units within a power plant by selecting appropriate monitoring equipment based on user monitoring request information, acquiring operating parameter data of each power generation unit in real time, and outputting real-time operating parameter data in reference to timestamp information. By acquiring and monitoring operating parameter data of power generation units in real time, the present invention enables timely detection of abnormal situations, provides advance warnings for possible failures, supports the optimization of control measures, improves the operating efficiency of the power plant, reduces downtime due to failures, enhances the reliability and safety of the equipment, and simultaneously contributes to energy saving and cost reduction, thereby realizing intelligent and optimized operation of the power plant.

[0039] In an intelligent optimization control method for a supercritical thermal power generation unit provided in an embodiment of the present invention, step 1 further includes extracting features from real-time operation parameter data, constructing and acquiring a first feature set, mapping and acquiring corresponding historical operation parameters from a historical database based on the first feature set, and outputting historical operation parameter data.

[0040] In this embodiment, the first feature set refers to a feature set extracted from real-time operation parameter data and is used to describe the operating state of the power generation unit (for example, features such as load magnitude, temperature change rate, and vibration frequency).

[0041] In this embodiment, the historical database refers to a database that stores historical operating parameter data and is used to store operating data of the power generation unit over a certain period in the past (for example, including load, temperature, and vibration data of the power generation unit over the past week).

[0042] In this embodiment, historical operation parameter data refers to operation parameter data of the power generation unit recorded within a certain period in the past, and is used to compare and analyze the operation conditions at different time periods (for example, the load of power generation unit A last week was 120 MW and the temperature was 280°C).

[0043] The implementation principle and beneficial effects of this embodiment are as follows. The present invention constructs a first feature set by extracting features from real-time operating parameter data, then maps corresponding historical operating parameter data from a historical database to obtain historical operating status data, which is used for analysis and comparison. By analyzing historical operating parameter data, the present invention can discover information such as operating trends and periodicity patterns in power generation units, which helps predict future potential problems, optimize control strategies, and improve the operating efficiency and reliability of the power plant. At the same time, by performing a comprehensive analysis using real-time and historical data as reference, it is possible to formulate more accurate control strategies, reduce the risk of failure, and contribute to improving the overall performance and economic benefits of the power plant.

[0044] Before outputting the initial data in a batch, the intelligent optimization control method for a supercritical thermal power generation unit provided in an embodiment of the present invention includes selecting and obtaining a preprocessing method to map from a method database based on monitoring request information and a first feature set, and performing preprocessing on real-time operation parameter data and historical operation parameter data based on the preprocessing method.

[0045] In this embodiment, the method database refers to a database that stores various preprocessing methods and is used to select an appropriate preprocessing method based on monitoring request information and feature sets (for example, preprocessing methods such as data cleaning, data smoothing, and data normalization).

[0046] In this embodiment, the preprocessing method refers to a method for preprocessing data, aimed at extracting useful information, reducing noise, or adjusting the data distribution (for example, it may include methods such as data smoothing, data dimensionality reduction, and outlier handling).

[0047] In this embodiment, preprocessing refers to the process of cleaning, transforming, or adjusting data for subsequent analysis and processing (for example, removing outliers and normalizing real-time operation parameter data, and smoothing historical operation parameter data).

[0048] The implementation principle and beneficial effects of this embodiment are as follows: The present invention selects an appropriate preprocessing method from a method database based on monitoring request information and a first feature set, and then uses these methods to preprocess real-time operating parameter data and historical operating parameter data to ensure data quality and applicability. By processing the data with the preprocessing method, the present invention can reduce noise and outliers in the data, making the data more accurate and reliable. This contributes to improving the accuracy of subsequent data analysis and modeling, optimizing the formulation of control policies, and further enhancing the operating efficiency and stability of the power plant. Preprocessing can also facilitate the understanding and application of data, providing stronger support for subsequent data processing and decision-making.

[0049] In the intelligent optimization control method for a supercritical thermal power generation unit provided in an embodiment of the present invention, selecting and acquiring a first model from a model database with reference to a selection command includes acquiring a user-inputted model manual selection command and a system-generated model automatic selection command, outputting a model selection command with reference to the priority corresponding to each command, and selecting and acquiring a first model to map from the model database based on the model selection command.

[0050] In this embodiment, the model selection command refers to a command manually entered by the user, which is used to guide the system to select a model appropriate to the current situation (for example, the user can select and specify a neural network-based model to predict load changes in a power generation unit).

[0051] In this embodiment, the automatic model selection command refers to a command automatically generated by the system based on certain rules or algorithms, and is used to select a model based on data characteristics or requirements (for example, the system can automatically select to use the ARIMA model for load prediction based on the time-series characteristics of the data).

[0052] In this embodiment, the model selection command refers to a command that takes into account the manual model selection command and the automatic model selection command, and outputs a final model selection command based on the corresponding priority of each command (for example, if the priority of the manual model selection command specified by the user is higher than that of the automatic model selection command, the system will prioritize the model specified by the user).

[0053] The implementation principle and beneficial effects of this embodiment are as follows. Based on the user-inputted model manual selection command and the system-generated model automatic selection command, the present invention outputs a final model selection command, taking into account the priority of each command. Subsequently, a first model to be mapped is selected from the model database based on this command. By taking into account the manual selection command and the system's automatic selection command, the present invention can fully consider the user's requirements and the system's intelligent selection, thereby ensuring that the optimal model for the current situation is selected. This can improve the accuracy and applicability of the model, further optimize control strategies, and enhance the operating efficiency and performance of the power plant. At the same time, such a method combining manual and automatic selection can increase the flexibility and intelligence of the system, bringing more convenience and benefits to the management and operation of the power plant.

[0054] In the intelligent optimization control method for a supercritical thermal power generation unit provided in an embodiment of the present invention, training and optimizing a first model using initial data includes performing data partitioning on the initial data using a predetermined data partitioning method to obtain a training dataset, a validation dataset, and a test dataset, and performing training and optimization on the first model based on the training dataset, the validation dataset, and the test dataset.

[0055] In this embodiment, the predetermined data splitting method refers to a pre-configured dataset splitting method used to divide the initial data into a training dataset, a validation dataset, and a test dataset. This contributes to evaluating the performance of the model on different datasets (for example, a common splitting ratio of 70%-15%-15% can be used to split the data into a training set, a validation set, and a test set).

[0056] In this embodiment, the training dataset refers to the dataset used to train the model, and the model learns its parameters through this data (for example, including historical operation parameter data and corresponding output results).

[0057] In this embodiment, the validation dataset refers to a dataset used to tune the model's hyperparameters and evaluate its performance. Validating the dataset helps to avoid the model fitting to the training data (for example, it can be used to adjust the neural network's learning rate or select the optimal model complexity).

[0058] In this embodiment, the test dataset refers to the dataset used to ultimately evaluate the model's performance. The model is tested on this dataset to evaluate its generalization ability (for example, it is used to simulate future real-time driving parameter data to verify the model's performance during actual operation).

[0059] The implementation principle and beneficial effects of this embodiment are as follows. The present invention divides an initial dataset into a training dataset, a validation dataset, and a test dataset based on a predetermined data partitioning method, and then trains and optimizes a first model based on these datasets. The training data is used to learn model parameters, the validation data is used to tune the model's hyperparameters, and the test data is used to finally evaluate the model's performance. By rationally partitioning the dataset and using it to train and optimize the model, the present invention can enhance the generalization ability and stability of the model. The training dataset is used for model learning, the validation dataset is used to tune the model to avoid overfitting, and the test dataset is used to finally evaluate the model's performance. This ensures that the model has relatively good predictive and generalization capabilities in actual applications, further improving the operating efficiency and control accuracy of the power plant.

[0060] In an intelligent optimization control method for a supercritical thermal power generation unit provided in an embodiment of the present invention, step 3 includes: extracting features from monitoring request information, constructing a second feature set based on the extracted features; obtaining corresponding indicator selection factors based on the second feature set and with reference to a predetermined feature-factor comparison table, constructing and obtaining a selection factor set; selecting and obtaining a first indicator to map to each indicator selection factor in the selection factor set from the indicator database with reference to a predetermined factor-indicator mapping table, constructing and obtaining an indicator candidate pool; obtaining an automatic indicator selection command generated by the system using a predetermined algorithm in real time, simultaneously obtaining a human-inputted indicator selection command via a predetermined port in real time, and outputting an indicator selection command set with reference to priority information corresponding to each command; and selecting and obtaining a predetermined evaluation indicator to map from the indicator candidate pool based on the indicator selection command set. This includes constructing and obtaining an evaluation index set, inputting the evaluation index set into a predetermined analysis model, performing setting adjustments and updates to the predetermined analysis model with reference to a predetermined self-adaptive algorithm, simultaneously outputting model format request data after the setting update, simultaneously obtaining corresponding data format information for all data included in the initial data and outputting an initial data format comparison table, selecting and obtaining a predetermined format conversion method from a method database to map the model format request information to the initial data format comparison table, performing format conversion on the initial data based on the model format request data and the initial data format comparison table, and using the predetermined format conversion method to output the data to be analyzed, and performing optimization analysis on the data to be analyzed through the predetermined analysis model with reference to the evaluation index set and after the setting update, and outputting the optimization analysis results.

[0061] In this embodiment, the second feature set refers to the feature set obtained by extracting features from the monitored demand information, and is used to build the foundation for evaluation metrics and optimization analysis (for example, features such as the slope of the load curve and frequency fluctuations extracted from real-time operating data of a thermal power generation unit).

[0062] In this embodiment, the predetermined feature-factor comparison table refers to a pre-configured comparison table that includes the mapping relationship between each feature in the second feature set and the indicator selection factor.

[0063] In this embodiment, the indicator selection factor refers to a feature factor obtained based on a predetermined feature-factor comparison table, and is used for the following indicator selection and construction of the evaluation indicator set.

[0064] In this embodiment, the selection factor set refers to a set composed of indicator selection factors.

[0065] In this embodiment, the predetermined factor-index mapping table refers to a pre-configured comparison table that includes the mapping relationship between the index selection factor and the evaluation index.

[0066] In this embodiment, the indicator database refers to a database containing various indicators for various power generation systems.

[0067] In this embodiment, the first indicator refers to an evaluation indicator selected from the indicator database.

[0068] In this embodiment, the candidate index pool refers to a set of candidate evaluation indices obtained by mapping based on factors within the selection factor set (for example, indices such as "load balance" and "frequency stability").

[0069] In this embodiment, the predetermined algorithm refers to an algorithm that has been set in advance for generating an automatic indicator selection command.

[0070] In this embodiment, the automatic indicator selection command refers to a command generated by a predetermined algorithm and used for the automatic selection of evaluation indicators (for example, a command in which the system automatically generates evaluation indicators based on real-time data characteristics).

[0071] In this embodiment, a predetermined port refers to a pre-configured interface or channel for data transmission and communication between the system and the user (for example, a dedicated port installed in the system to receive user input commands).

[0072] In this embodiment, an artificial indicator selection command refers to a command entered by a human to guide the system to select a specific indicator or parameter (for example, a user may instruct the system to prioritize certain critical operating indicators).

[0073] In this embodiment, priority information refers to information used to indicate the importance or priority of different commands or operations (for example, commands with higher priority are executed preferentially).

[0074] In this embodiment, the indicator selection command set refers to a set of commands generated by the system to select a specific indicator or factor, including manual indicator selection commands and automatic selection commands (for example, a set of commands generated by the system based on user input and an algorithm).

[0075] In this embodiment, the evaluation index set refers to a set of indicators used to evaluate system performance or results (for example, including indicators such as power plant efficiency and load balance).

[0076] In this embodiment, a predetermined self-adaptive algorithm refers to a pre-configured algorithm that automatically adjusts parameters or settings based on system conditions or requirements (for example, the system may automatically adjust the parameters of a control algorithm based on real-time data).

[0077] In this embodiment, setting adjustment and updating refer to operations that adjust and update system settings based on evaluation results or requests (for example, adjusting and controlling policies based on evaluation metric results).

[0078] In this embodiment, the model format request data refers to data in a specific format required by the model (for example, a neural network model may input data of a certain dimension and range).

[0079] In this embodiment, data format information refers to information indicating the data structure and type (for example, information indicating whether the data is time-series data or classification data).

[0080] In this embodiment, the initial data format comparison table refers to a comparison table that records the correspondence between the format information of the initial data and the format information required by the model.

[0081] In this embodiment, the predetermined format conversion method refers to a method pre-configured to convert the initial data into a format required by the model (for example, converting time-series data into a matrix format).

[0082] In this embodiment, the data to be analyzed refers to data that has undergone format conversion before being used for analysis (for example, data that has been processed using a predetermined format conversion method).

[0083] The implementation principle and beneficial effects of this embodiment are as follows: This invention constructs an evaluation index set through features extraction, factor selection, and index matching, and performs optimization analysis on the data to be analyzed by adjusting and updating settings using a predetermined analysis model and self-adaptive algorithm. By referring to the selection of human and automatic commands, it achieves automatic adaptation to the system state while better meeting user requirements. At the same time, through format conversion and data optimization analysis, it improves the accuracy of the analysis results and the level of system intelligence, further optimizing the operating efficiency and control accuracy of the power plant. As a result, the system can more effectively respond to complex control requests and data processing, increasing the overall performance and efficiency of the system.

[0084] In an intelligent optimization control method for a supercritical thermal power generation unit provided in an embodiment of the present invention, step 4 involves inputting predetermined operating threshold conditions into a predetermined analysis model to generate predetermined boundary conditions, acquiring self-generated optimization indicators output by the predetermined analysis model based on the predetermined boundary conditions and artificially selected optimization indicators, outputting an optimization indicator set based on the self-generated optimization indicators and artificially selected optimization indicators, inputting initial data into a predetermined analysis model to perform predictive analysis, acquiring a first prediction result by a variable structure predictive control method and a second prediction result by a generalized predictive control method, creating a process based on the optimization indicator set, the first prediction result, and the second prediction result, and referring to a predetermined policy-method, creating and acquiring an intelligent control policy and an intelligent control method, and the optimization indicator set, the intelligent control policy and intelligent This includes constructing and obtaining an index-policy-method comparison table based on the mapping relationship with the control method; extracting features from the optimization analysis results and constructing and obtaining an optimization feature set based on the extracted features; obtaining optimization indices that map to each feature in the optimization feature set by referring to a predetermined feature-index mapping table and outputting an optimization index set awaiting execution; obtaining intelligent control policies and intelligent control methods that map to each optimization index in the optimization index set awaiting execution by referring to the index-policy-method comparison table and outputting an execution policy-method list; and performing control and adjustment on the corresponding operating parameters in each power generation unit based on the intelligent control policies and intelligent control methods in the execution policy-method list, and generating process data in real time and feeding it back to a predetermined data analysis model.

[0085] In this embodiment, the predetermined operating threshold condition refers to a pre-set condition used to guide the operating threshold under specific system conditions (for example, setting the generator load factor not to exceed 90%).

[0086] In this embodiment, the predetermined boundary conditions refer to conditions generated based on predetermined operating threshold conditions and are used to define the boundaries of system operation (for example, boundary conditions set based on load factor).

[0087] In this embodiment, the self-generated optimization index refers to an index that is automatically generated for optimization and output from a predetermined analysis model (for example, an efficiency index generated by the system based on data).

[0088] In this embodiment, the artificially optimized metrics refer to metrics for optimization that have been artificially selected (for example, key performance metrics selected by the operator based on experience).

[0089] In this embodiment, the optimization metric set refers to a set of metrics that include self-generated optimization metrics and artificial optimization metrics (for example, metrics such as efficiency and load balance).

[0090] In this embodiment, the variable structure predictive control method refers to a model-based control method that optimizes system performance by predicting the operation of the system over a certain period of time in the future using a system model within each control cycle and optimizing the control input (for example, dynamically adjusting the control strategy based on real-time data).

[0091] In this embodiment, the first prediction result refers to the first prediction result obtained under the variable structure prediction control method (for example, predicting load changes over the next hour).

[0092] In this embodiment, the generalized predictive control method refers to a prediction-based control method that achieves optimized control of a system by using a system model to predict the system response over a certain period of time in the future and adjusting the control input based on these prediction results (for example, performing real-time calculations and predictions using a recursive algorithm).

[0093] In this embodiment, the second prediction result refers to the second prediction result obtained under the generalized prediction control method (for example, predicting the power grid load demand for the following day).

[0094] In this embodiment, the predetermined policy-method formulation process refers to the process of formulating intelligent control policies and methods (including, for example, steps such as determining optimization targets, selecting control policies, and formulating control methods).

[0095] In this embodiment, the intelligent control strategy and intelligent control method refer to a strategy and method for optimizing the performance of the control system, where the strategy specifies the overall direction of control, and the method specifies specific control means and algorithms.

[0096] In this embodiment, the index-policy-method comparison table refers to a table that records the mapping relationship between optimization indicators and intelligent control policies and intelligent control methods (for example, specifying that a particular control policy and control method will be adopted under a certain optimization indicator).

[0097] In this embodiment, the optimization index set refers to a set of features extracted from the optimization analysis results, which are used to guide the selection of optimization indexes. These features can reflect the system's performance, state, or other important information.

[0098] In this embodiment, the predetermined feature-index mapping table refers to a table containing the mapping relationship between each feature in the optimization feature set and the optimization index, and is used to select an optimization index to match based on the optimization feature set.

[0099] In this embodiment, the set of optimization metrics awaiting execution refers to the set of optimization metrics awaiting execution determined based on the optimization feature set and the mapping table, and these metrics are used to determine the intelligent control policy and the intelligent control method.

[0100] In this embodiment, the execution-waiting policy-method list refers to the execution-waiting intelligent control policy and method list determined based on the indicator-policy-method comparison table, and these policies and methods are applied to the system to perform real-time control.

[0101] In this embodiment, process data refers to real-time data generated during the control process, which is used to feed back into a predetermined data analysis model for further analysis and adjustment.

[0102] The implementation principle and beneficial effects of this embodiment are as follows. The present invention generates predetermined boundary conditions based on predetermined operating threshold conditions, and then generates an optimization index set based on automatically generated and manually selected optimization indexes. Next, intelligent control strategies and methods are formulated using the results obtained from predictive analysis and the optimization index set, and an index-strategy-method comparison table is constructed. Finally, features are extracted based on the optimization analysis results, an optimization feature set is generated, corresponding optimization indexes are selected, and the intelligent control strategies and methods are implemented. By combining automatically generated indexes and manually selected indexes, the present invention enables the system to evaluate system performance more comprehensively and formulate intelligent control strategies and methods based on predictive results. This method helps the system to more accurately adjust operating parameters, improve system stability and efficiency, and further optimize the operation and control effects of the power plant. At the same time, by constructing an index-strategy-method comparison table, the system can be made more intelligent and adaptive, improving the overall performance and responsiveness of the system.

[0103] Finally, it should be noted that the following embodiments are merely for illustrating, and not limiting, the technical solutions of the present invention. Although the present invention has been described in detail with reference to the embodiments described above, those skilled in the art will understand that modifications can still be made to the technical solutions described in the embodiments above, or that some of the technical features therein can be substituted with equivalents. However, these modifications or substitutions will not cause the essence of the relevant technical solutions to deviate from the spirit and scope of the technical solutions relating to each embodiment of the present invention.

Claims

1. An intelligent optimization control method for a supercritical thermal power generation unit, Including steps 1 through 4, Here, step 1 involves acquiring real-time operating parameters and historical operating parameters for each power generation unit within the power plant, and outputting the initial data together. The acquisition of real-time operating parameters for each power generation unit within a power plant includes acquiring user monitoring request information, selecting and acquiring appropriate predetermined monitoring equipment from an equipment database based on the monitoring request information, and acquiring operating parameter data for each power generation unit in real time based on the predetermined monitoring equipment, and outputting the real-time operating parameter data with reference to the timestamp corresponding to the operating parameter data. Step 2 involves selecting and acquiring a first model from the model database based on the selection command, training and optimizing the first model using the initial data, and outputting a predetermined analysis model. Step 3 involves performing an optimization analysis on the initial data based on the predetermined analysis model and with reference to predetermined evaluation indicators, and outputting the optimization analysis results. Step 3 above is, The process involves extracting features from the aforementioned monitoring request information and constructing a second feature set based on the extracted features. Based on the aforementioned second set of features, and with reference to a predetermined feature-factor comparison table, the corresponding indicator selection factors are obtained, and a set of selection factors is constructed and obtained. Referencing a predetermined factor-indicator mapping table, select and obtain a first indicator from the indicator database to map to each indicator selection factor within the aforementioned selection factor set, and construct and obtain an indicator candidate pool. The system acquires automatic indicator selection commands generated by a predetermined algorithm in real time, simultaneously acquires human-inputted indicator selection commands via a predetermined port in real time, and outputs an indicator selection command set by referring to priority information corresponding to each command. Based on the aforementioned indicator selection command set, a predetermined evaluation indicator to be mapped is selected and obtained from the indicator candidate pool, and an evaluation indicator set is constructed and obtained. The evaluation indicator set is input into a predetermined analysis model, and the predetermined analysis model is adjusted and updated with reference to a predetermined self-adaptive algorithm, and at the same time, data for the model format request after the setting update is output. Simultaneously, the system obtains the corresponding data format information for all data included in the initial data and outputs an initial data format comparison table. The method involves selecting and obtaining a predetermined format conversion method from the database that maps the model format request information to the initial data format comparison table, Based on the data of the model format request and the initial data format comparison table, and using the predetermined format conversion method, the system performs a format conversion on the initial data and outputs the data to be analyzed. This further includes performing an optimization analysis on the data to be analyzed using the aforementioned set of evaluation indicators and the predetermined analysis model after the settings have been updated, and outputting the optimization analysis results. Step 4 involves extracting features from the optimization analysis results, selecting and obtaining control strategies and control methods that map to features extracted from the strategy database with reference to a predetermined feature strategy comparison table, and adjusting the operating parameters of each power generation unit based on the control strategies and control methods. An intelligent optimization control method for a supercritical thermal power generation unit.

2. Step 1 further includes extracting features from the real-time operation parameter data, constructing and acquiring a first feature set, mapping and acquiring corresponding past historical operation parameters from a historical database based on the first feature set, and outputting past historical operation parameter data. The intelligent optimization control method for a supercritical thermal power generation unit as described in claim 1.

3. Before outputting the initial data all at once, the intelligent optimization control method for the supercritical thermal power generation unit is as follows: Based on the aforementioned monitoring request information and the first feature set, a preprocessing method to be mapped is selected and obtained from the method database. This includes performing preprocessing on the real-time operation parameter data and the historical operation parameter data based on the preprocessing method described above. The intelligent optimization control method for a supercritical thermal power generation unit according to claim 2.

4. The above-mentioned process involves selecting and obtaining the first model from the model database based on the selection command. The system acquires the user-inputted model selection command and the system-generated automatic model selection command, and outputs a model selection command based on the priority of each command. This includes selecting and obtaining a first model to map from a model database based on the aforementioned model selection command. The intelligent optimization control method for a supercritical thermal power generation unit as described in claim 1.

5. The above-mentioned method of training and optimizing the first model using the initial data is as follows: The initial data is divided using a predetermined data division method to obtain a training dataset, a validation dataset, and a test dataset. This includes training and optimizing the first model based on the training dataset, the validation dataset, and the test dataset. The intelligent optimization control method for a supercritical thermal power generation unit as described in claim 1.

6. Step 4 is, Inputting predetermined operating threshold conditions into the predetermined analysis model and generating predetermined boundary conditions, Based on the predetermined boundary conditions, the system obtains self-generated optimization indicators output by a predetermined analysis model and artificial optimization indicators obtained through artificial selection, and outputs an optimization indicator set based on the self-generated optimization indicators and the artificial optimization indicators. The initial data is input into the predetermined analysis model to perform predictive analysis, and a first prediction result obtained by the variable structure predictive control method and a second prediction result obtained by the generalized predictive control method are obtained. Based on the aforementioned optimization index set, the first prediction result, and the second prediction result, and with reference to a predetermined policy-method, a process is created, an intelligent control policy and an intelligent control method are created and obtained, and an index-policy-method comparison table is constructed and obtained based on the mapping relationship between the aforementioned optimization index set, the intelligent control policy, and the intelligent control method. The process involves extracting features from the aforementioned optimization analysis results, constructing and obtaining an optimized feature set based on the extracted features, and Referencing a predetermined feature-index mapping table, obtain the optimization index to map to each feature in the optimization feature set, and output the optimization index set awaiting execution. Referencing the aforementioned indicator-policy-method comparison table, obtain the intelligent control policy and intelligent control method that map to each optimization indicator in the pending optimization indicator set, and output the pending policy-method list. This includes controlling and adjusting the corresponding operating parameters within each power generation unit based on the intelligent control strategies and intelligent control methods in the aforementioned execution waiting strategy-method list, and generating process data in real time and feeding it back to a predetermined data analysis model. The intelligent optimization control method for a supercritical thermal power generation unit as described in claim 1.

Citation Information

Patent Citations

  • Self-adaptive control method for main steam temperature of supercritical thermal power generating unit and related device

    CN117518823A

  • Thermal power generating unit operation optimization method and system based on energy efficiency optimization

    CN119093489A