Power system large-scale unit combination problem optimization method and related equipment

Abstract

The embodiment of the invention belongs to the field of power systems and automation thereof, and relates to a power system large-scale unit commitment problem optimization method, which comprises the following steps: acquiring historical data of power system case clearing; performing feature extraction and preprocessing on the historical data to obtain feature data; according to a preset condition, performing problem difficulty grade labeling on the feature data to obtain a labeled data set; constructing a problem difficulty prediction model according to the labeled data set and a preset machine learning algorithm; according to the problem difficulty prediction model, dynamic adjustment of solver parameters of the large-scale unit commitment problem is achieved. The invention further provides a power system large-scale unit combination problem optimization device, computer equipment and a storage medium. According to the method, machine learning is combined on the basis of operation planning optimization, the solving efficiency of the large-scale unit commitment problem is improved, and the optimization quality of the large-scale unit commitment problem is improved.

Description

Technical Field

[0001] The present application relates to the field of power systems and automation thereof, and in particular to a method for optimizing large-scale unit combination problems in power systems and related equipment. Background Art

[0002] Power system market clearing refers to the process of achieving power supply and demand balance and electricity price determination through market mechanisms while ensuring the stable operation of the power system. In this process, the most critical link is the large-scale unit combination problem. Its core goal is to minimize the total operating cost of the system under the conditions of meeting various constraints. This cost usually includes the start-up and shutdown costs of the generator sets, operating costs, system standby costs, and power transmission costs under system constraints.

[0003] Generally speaking, in order to ensure the reliability of the power system, large-scale unit combination problems must meet a series of complex constraints, including unit start-stop constraints, ramp capacity constraints, output range constraints, system reserve constraints, and supply-demand balance, etc. However, with the expansion of the scale of the power system, especially the continuous increase in the proportion of renewable energy power generation, the complexity and difficulty of solving large-scale unit combination problems have also increased significantly. Actual large-scale unit combination problems often contain millions of variables and constraints, and have strong coupling and nonlinear characteristics.

[0004] At present, the traditional solution to the unit commitment problem is based on the mixed integer linear programming (MILP) method, and its related algorithms include branch and bound method, cutting plane method and dynamic programming method. However, it is difficult to quickly solve large-scale problems using the mixed integer linear programming (MILP) method, resulting in low solution efficiency. In addition, due to the different daily load demands of the power grid, the status of the generator sets and the uncertainty of renewable energy generation, the difficulty of large-scale unit commitment problems also changes dynamically. Problems with higher difficulty often require longer solution time. It is difficult to cope with complex and changing problem scenarios using only this solution method, resulting in poor optimization quality of large-scale unit commitment problems. Summary of the invention

[0005] The purpose of the embodiments of the present application is to propose a method for optimizing the large-scale unit combination problem of a power system and related equipment to solve the technical problems of low solution efficiency and poor optimization quality of the large-scale unit combination problem.

[0006] In order to solve the above technical problems, the embodiment of the present application provides a method for optimizing the large-scale unit combination problem of a power system, which adopts the following technical solutions:

[0007] A method for optimizing large-scale unit commitment problems in a power system comprises the following steps:

[0008] Obtain historical data on power system clearing cases;

[0009] Extracting and preprocessing the historical data to obtain feature data;

[0010] According to preset conditions, the feature data is labeled with a difficulty level of the question to obtain a labeled data set;

[0011] Constructing a problem difficulty prediction model based on the labeled data set and a preset machine learning algorithm;

[0012] According to the problem difficulty prediction model, dynamic adjustment of solver parameters for large-scale unit commitment problems is achieved.

[0013] Furthermore, the step of dynamically adjusting the solver parameters of the large-scale unit commitment problem according to the problem difficulty prediction model specifically includes:

[0014] Inputting the unit commitment problem to be predicted into the problem difficulty prediction model;

[0015] Outputting the problem difficulty level corresponding to the unit combination problem according to the problem difficulty prediction model;

[0016] The solver parameters are set according to the difficulty level of the problem.

[0017] Furthermore, the difficulty level of the problem includes simple difficulty, medium difficulty and difficult difficulty, and the solver parameters include the maximum number of iterations, convergence accuracy, relative error tolerance, number of parallel processing threads, heuristic algorithm switch and initial slack variable. The step of setting the solver parameters according to the difficulty level of the problem specifically includes:

[0018] If the difficulty level of the problem is simple, the maximum number of iterations is set to the first number, the convergence accuracy is set to the first accuracy, the relative error allowable range is set to the first range, the number of parallel processing threads is set to the first value, the heuristic algorithm switch is set to off, and the initial slack variable is set to a larger value;

[0019] If the difficulty level of the problem is medium difficulty, the maximum number of iterations is set to the second number, the convergence accuracy is set to the second accuracy, the relative error allowable range is set to the second range, the number of parallel processing threads is set to the second value, the heuristic algorithm switch is set to on, and the initial slack variable is set to a medium value;

[0020] If the difficulty level of the problem is difficult, the maximum number of iterations is set to the third number, the convergence accuracy is set to the third accuracy, the relative error allowable range is set to the third range, the number of parallel processing threads is set to the third value, the heuristic algorithm switch is set to on, and the initial slack variable is set to a smaller value;

[0021] The second number is greater than the first number and less than the third number, the second precision is less than the first precision and greater than the third precision, the second range is less than the first range and greater than the third range, and the second value is greater than the first value and less than the third value.

[0022] Furthermore, the step of labeling the feature data with the difficulty level of the questions according to the preset conditions to obtain the labeled data set specifically includes:

[0023] analyzing the solution time of historical unit commitment problems in the characteristic data;

[0024] If the solution time is less than the first time length, the historical unit commitment problem is marked as a simple difficulty;

[0025] If the solution time is greater than or equal to the first time length and less than or equal to the second time length, the historical unit commitment problem is marked as medium difficulty;

[0026] If the solution time is greater than the second time length, marking the historical unit commitment problem as difficult;

[0027] The labeled feature data is used as the labeled data set.

[0028] Furthermore, the step of constructing a question difficulty prediction model based on the labeled data set and a preset machine learning algorithm specifically includes:

[0029] Dividing the labeled data set into a training set and a test set;

[0030] According to the training set and the machine learning algorithm, an initial machine learning model is established and model training is performed to obtain a trained machine learning model;

[0031] The machine learning model is tested according to the test set, and parameters of the machine learning model are optimized according to the test results to obtain the problem difficulty prediction model.

[0032] Furthermore, the step of testing the machine learning model according to the test set and optimizing the parameters of the machine learning model according to the test result to obtain the problem difficulty prediction model specifically includes:

[0033] Inputting the test set into the machine learning model to obtain an output result;

[0034] Verify and evaluate the output result according to the historical data to obtain the test result;

[0035] generating a model adjustment strategy based on the test results;

[0036] The parameters of the machine learning model are optimized according to the model adjustment strategy, and the machine learning model after the parameter optimization is used as the problem difficulty prediction model.

[0037] Furthermore, the step of extracting and preprocessing the historical data to obtain feature data specifically includes:

[0038] Extracting features from the historical data to obtain a feature vector set;

[0039] The feature vector set is subjected to missing value processing, outlier processing, data standardization and data dimension reduction to obtain the feature data.

[0040] In order to solve the above technical problems, the embodiment of the present application also provides a device for optimizing large-scale unit combination problems in a power system, which adopts the following technical solution:

[0041] A device for optimizing large-scale unit combination problems in a power system, comprising:

[0042] An acquisition module, used to obtain historical data of power system clearing cases;

[0043] A processing module, used for performing feature extraction and preprocessing on the historical data to obtain feature data;

[0044] A labeling module, used to label the feature data with a difficulty level according to preset conditions to obtain a labeled data set;

[0045] A construction module, used to construct a question difficulty prediction model based on the labeled data set and a preset machine learning algorithm;

[0046] The implementation module is used to dynamically adjust the solver parameters of the large-scale unit commitment problem according to the problem difficulty prediction model.

[0047] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0048] A computer device comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the above-mentioned method for optimizing large-scale unit combination problems in a power system are implemented.

[0049] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0050] A computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the above-mentioned method for optimizing large-scale unit combination problems in a power system.

[0051] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0052] The optimization method for large-scale unit combination problems in power systems disclosed in the present application obtains historical data of power system clearing cases; then extracts and preprocesses the historical data to obtain feature data; then, based on preset conditions, labels the feature data with a problem difficulty level to obtain a labeled data set; then, based on the labeled data set and a preset machine learning algorithm, a problem difficulty prediction model is constructed; thereby, according to the problem difficulty prediction model, dynamic adjustment of solver parameters for large-scale unit combination problems is achieved. The present application effectively reduces the waste of computing resources when solving large-scale unit combination problems by combining machine learning with operations research optimization, ensures the stability of power system operation, and improves the efficiency of solving large-scale unit combination problems. In addition, it provides an innovative solution for power system market clearing and improves the optimization quality of large-scale unit combination problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0054] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0055] Figure 2 is a flow chart of an embodiment of a method for optimizing a large-scale unit combination problem in a power system according to the present application;

[0056] Figure 3It is a flowchart for constructing a problem difficulty prediction model training according to an embodiment of a method for optimizing a large-scale unit commitment problem in a power system of the present application;

[0057] Figure 4 is a flow chart of dynamic adjustment of solver parameters according to an embodiment of the method for optimizing large-scale unit commitment problem of power system of the present application;

[0058] Figure 5 It is a structural schematic diagram of an embodiment of a device for optimizing large-scale unit combination problems in a power system according to the present application;

[0059] Figure 6 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0060] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0061] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0062] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0063] like Figure 1 As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0064] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0065] Terminal devices 101, 102, 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 (Moving Picture Experts Group Audio Layer III) players, MP4 (Moving Picture Experts Group Audio Layer IV) players, laptop computers, desktop computers, etc.

[0066] The server 105 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal devices 101 , 102 , and 103 .

[0067] It should be noted that the method for optimizing the large-scale unit combination problem of a power system provided in the embodiment of the present application is generally executed by a server, and accordingly, the device for optimizing the large-scale unit combination problem of a power system is generally arranged in the server.

[0068] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0069] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for optimizing a large-scale unit combination problem in a power system according to the present application. The method for optimizing a large-scale unit combination problem in a power system comprises the following steps:

[0070] Step S201, obtaining historical data of power system clearing cases.

[0071] In this embodiment, the optimization method for large-scale unit commitment problem of power system is executed on the electronic device (e.g. Figure 1The server shown in the figure) can send or receive data through a wired connection or a wireless connection. It should be noted that the above wireless connection method may include but is not limited to 3G / 4G / 5G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other wireless connection methods currently known or to be developed in the future.

[0072] In this embodiment, in the operation of the power market, the clearing case records the details of the balance of power supply and demand, including the start and stop time of the generator set, load demand, market clearing price, etc. These historical data are obtained from the power market database through an automated interface to ensure the integrity and accuracy of the data. Specifically, the power market will clear the market every day based on the load demand of the day and the quotations of the generator sets. The system will regularly extract historical data from the power market system, which includes the quotation of each generator set, the cleared electricity, the market price, and the related start and stop status. By obtaining a large amount of historical data, the system can better understand the past power market clearing situation, which provides a basis for subsequent feature extraction and machine learning model training.

[0073] Step S202: extract and preprocess the historical data to obtain feature data.

[0074] In this embodiment, key features related to the unit combination problem are extracted from the acquired historical data, mainly including the quotation, start and stop time, load demand, etc. of the generator set. The data needs to be preprocessed before use, such as removing missing values, normalization, and outlier processing. For example, assuming that the historical data contains information such as the clearing quotation and market load of the generator set, the system first extracts the quotation data, load demand, start and stop status, etc. of each unit, normalizes these data so that their values ​​can be compared and analyzed within a unified range, and processes any missing or abnormal data. Through feature extraction and preprocessing, the original complex historical data is converted into structured feature data, providing high-quality input for subsequent machine learning model training.

[0075] First, feature extraction is performed on the historical data of power market clearing cases. Some data features are shown in Table 1:

[0076] Table 1 Some data characteristics

[0077]

[0078]

[0079] This table shows the different features of system data, power plant data, and unit data and their corresponding sub-tables. The features include power, output, spare capacity, ramp rate, etc. After extracting the features, the feature data is preprocessed, including missing value processing, outlier processing, data standardization, data dimensionality reduction, etc., to improve data quality and ensure a high-quality data foundation for machine learning model training. This step is to prepare data for subsequent model training to ensure that the model can effectively learn and predict based on high-quality data.

[0080] Step S203: label the feature data with the difficulty level of the questions according to the preset conditions to obtain a labeled data set.

[0081] In this embodiment, the difficulty level of each problem is marked according to the difficulty of solving the problem in the historical data (for example, according to the solution time or resource usage). The difficulty levels are usually divided into simple, medium, and difficult, which is convenient for subsequent classification processing. For example, it is assumed that the solution time of the unit combination problem in the historical data is the key judgment criterion. The system analyzes the solution time of each problem in the past. If the solution time is less than 500 seconds, it is marked as simple difficulty; the solution time between 500 and 1500 seconds is marked as medium difficulty; and the solution time greater than 1500 seconds is marked as difficult. By marking the difficulty of the data, the data can be classified for model training of different difficulty levels, improving the generalization ability of the model and ensuring the adaptability of the model in different difficulty scenarios.

[0082] Step S204: construct a question difficulty prediction model based on the labeled data set and a preset machine learning algorithm.

[0083] In this embodiment, the labeled feature data is used to train with machine learning algorithms (such as random forests, neural networks, support vector machines, etc.) to build a problem difficulty prediction model, which can predict the difficulty level of new problems based on the input feature data. Specifically, the random forest algorithm is used to divide the labeled data set into a training set and a test set for model training. The model input is a processed feature vector, and the output is the difficulty level of the problem (simple, medium or difficult); after the training is completed, the test set is used to evaluate and tune the model; after the difficulty prediction model is built, the system can quickly predict the difficulty level of new problems, and provide a reference for the subsequent solution process, thereby improving the intelligence of the system.

[0084] like Figure 3As shown in the figure, first, collect the data source for machine learning, here is the "cleared case data", extract features useful for the model from the original data, and facilitate subsequent processing and model training, and then preprocess the data after feature extraction. Specifically, fill or process missing values ​​in the data to ensure data integrity; standardize the data to ensure that the data is analyzed on the same scale; identify and process outliers to avoid the negative impact of abnormal data on the model; reduce data dimensions through feature selection or dimensionality reduction algorithms to retain important information. Then classify the preprocessed data according to the difficulty of the problem so that it can be optimized for problems of different difficulty. Next, divide the training set and the test set, the training set is used for model training, and the test set is used for model verification. Then, use the training set to train the machine learning model, and then use the test set to verify and test the trained model to evaluate the performance of the model. The entire process covers the complete steps from data collection, feature extraction, data preprocessing to model training and verification, and is suitable for the standard workflow in machine learning or deep learning tasks.

[0085] Step S205, dynamically adjusting the solver parameters of the large-scale unit commitment problem according to the problem difficulty prediction model.

[0086] In this embodiment, the parameters of the solver are dynamically adjusted according to the predicted difficulty level. For simple problems, the number of iterations and the input of computing resources are reduced; for medium or difficult problems, the number of parallel threads is increased and the convergence accuracy is improved to ensure the optimal solution. Exemplarily, if the problem difficulty is predicted to be "simple", the system sets the maximum number of iterations of the solver to 500, the convergence accuracy to 1e-4, and turns off the heuristic algorithm. If the prediction is "difficult", the maximum number of iterations is increased to 5000, 8 parallel thread processing is enabled, and the heuristic algorithm is turned on. By dynamically adjusting the solver parameters, the system can flexibly allocate computing resources and time according to problems of different difficulty levels, avoiding waste of resources while ensuring high-quality solutions to complex problems.

[0087] This application obtains historical data of power system clearing cases, extracts features related to unit combination problems and performs preprocessing, and combines machine learning algorithms to predict the difficulty of the problem. While improving the efficiency of solving, it significantly improves the optimization quality of large-scale unit combination problems, ensures the efficient use of system computing resources, reduces redundant computing and resource waste, and thus optimizes the operational stability of the power system.

[0088] In some optional implementations of this embodiment, the step of dynamically adjusting the solver parameters of the large-scale unit commitment problem according to the problem difficulty prediction model specifically includes:

[0089] Inputting the unit commitment problem to be predicted into the problem difficulty prediction model;

[0090] Outputting the problem difficulty level corresponding to the unit combination problem according to the problem difficulty prediction model;

[0091] The solver parameters are set according to the difficulty level of the problem.

[0092] In this embodiment, the system first obtains the characteristic data of the unit combination problem that needs to be optimized, including the quotation, load demand, system constraints, etc. of the generator set. These characteristic data are input into the trained problem difficulty prediction model. Assuming that the unit combination problem that needs to be optimized currently involves the quotation and load demand of multiple generator sets, the system will organize these data into feature vectors, and then input them into the random forest model or neural network model for difficulty prediction. The problem difficulty prediction model outputs the difficulty level of the unit combination problem through calculation and analysis. The difficulty level is generally divided into simple, medium or difficult, and is classified based on factors such as the complexity of the problem and the solution time. If the model analyzes and believes that the characteristics of the current problem are similar to those of the difficult problems in the past, the "difficult" difficulty level will be output; if the problem is relatively simple in the feature dimension, the model may output the "simple" difficulty level, and the system automatically adjusts the parameter configuration of the solver according to the predicted difficulty level. Simple problems use fewer iterations and lower precision requirements, while complex problems increase the number of parallel processing threads, improve convergence accuracy and iterations to ensure that high-quality solutions are obtained within a reasonable time. If the problem difficulty is easy, set the maximum number of iterations to 500, the convergence accuracy to 1e-4, and turn off the heuristic algorithm to increase the solution speed; if the problem difficulty is medium, set the maximum number of iterations to 1000, enable 4 parallel processing threads, and turn on the heuristic algorithm to balance the solution speed and accuracy; if the problem difficulty is difficult, set the maximum number of iterations to 5000, increase the convergence accuracy to 1e-6, enable 8 parallel threads, and turn on the heuristic algorithm to ensure that the optimal solution is obtained within a reasonable time.

[0093] like Figure 4 As shown in the figure, first, case data for model training and prediction are collected, key features are extracted from the case data, and these features are important inputs for the model to predict the difficulty of the problem. The extracted feature data is preprocessed to ensure the quality and consistency of the data, including cleaning, normalization and other steps; then, the difficulty level of the current problem is predicted using the trained machine learning model, and the difficulty of the problem is classified according to the prediction results of the model, judging whether it is simple, medium or difficult; then, different solutions are set for problems of different difficulty, for example, parameter combination 1 is selected to solve these simpler problems; for problems of medium difficulty, parameter combination 2 is selected for processing; and parameter combination 3 is used to process more complex and difficult problems. Finally, the final solution is obtained based on the solutions to problems of different difficulty.

[0094] This application inputs the problem to be processed into a difficulty prediction model and automatically adjusts the solver parameters according to the predicted difficulty level. This step ensures that the system can reasonably allocate computing resources when solving large-scale unit combination problems, improve solution efficiency, and provide sufficient computing power to obtain accurate solutions for complex problems.

[0095] In some optional implementations of this embodiment, the difficulty level of the problem includes simple difficulty, medium difficulty and difficult difficulty, the solver parameters include the maximum number of iterations, convergence accuracy, relative error allowable range, number of parallel processing threads, heuristic algorithm switch and initial slack variable, and the step of setting the solver parameters according to the problem difficulty level specifically includes:

[0096] If the difficulty level of the problem is simple, the maximum number of iterations is set to the first number, the convergence accuracy is set to the first accuracy, the relative error allowable range is set to the first range, the number of parallel processing threads is set to the first value, the heuristic algorithm switch is set to off, and the initial slack variable is set to a larger value;

[0097] If the difficulty level of the problem is medium difficulty, the maximum number of iterations is set to the second number, the convergence accuracy is set to the second accuracy, the relative error allowable range is set to the second range, the number of parallel processing threads is set to the second value, the heuristic algorithm switch is set to on, and the initial slack variable is set to a medium value;

[0098] If the difficulty level of the problem is difficult, the maximum number of iterations is set to the third number, the convergence accuracy is set to the third accuracy, the relative error allowable range is set to the third range, the number of parallel processing threads is set to the third value, the heuristic algorithm switch is set to on, and the initial slack variable is set to a smaller value;

[0099] The second number is greater than the first number and less than the third number, the second precision is less than the first precision and greater than the third precision, the second range is less than the first range and greater than the third range, and the second value is greater than the first value and less than the third value.

[0100] In this embodiment, different solver parameters can be set for unit combination problems of different difficulty levels before solving. The solver parameters include the maximum number of iterations, convergence accuracy, relative error allowable range, number of parallel processing threads, heuristic algorithm switch and initial slack variable, among which the maximum number of iterations can be set to the first number, the second number, and the third number; the convergence accuracy can be set to the first accuracy, the second accuracy, and the third accuracy; the relative error allowable range can be set to the first range, the second range, and the third range; the number of parallel processing threads can be set to the first value, the second value, and the third value; the heuristic algorithm switch can be set to on or off; the initial slack variable can be set to a larger value, a medium value, or a smaller value. It should be noted that the second number is greater than the first number and less than the third number, the second accuracy is less than the first accuracy and greater than the third accuracy, the second range is less than the first range and greater than the third range, and the second value is greater than the first value and less than the third value. For simple problems, due to the low complexity of the problem, the solution speed can be accelerated by reducing computing resources and reducing the solution accuracy to reduce unnecessary computing overhead. For example, the maximum number of iterations is set to 500, and iterations are reduced to save computing time; the convergence accuracy is set to 1e-4, allowing a larger error to speed up the convergence speed; the relative error allowable range is set to 1%, and the error range is appropriately relaxed; the number of parallel processing threads is set to 1, and single-thread processing can meet the computing requirements; the heuristic algorithm switch is set to off to simplify the solution process; the initial slack variable is set to a larger value, allowing looser initial conditions to quickly obtain a preliminary solution. By reducing the number of iterations, accuracy, and number of threads, the system can quickly solve simple problems and avoid wasting resources. For medium-difficulty problems, more computing resources and precision are needed to ensure the quality of the solution, while ensuring that the solution time is within a reasonable range. For example, the maximum number of iterations is set to 1000 to ensure the accuracy and stability of the solution; the convergence precision is set to 1e-5, and the precision requirement is moderate; the relative error allowable range is set to 0.5%, allowing a small error to ensure that the result is close to the optimal, the number of parallel processing threads is set to 4, parallel processing is enabled to speed up the calculation, the heuristic algorithm is set to on, and heuristic methods are used to help speed up the solution of complex problems. The initial slack variable is set to a medium value to adapt to the complexity of medium-difficulty problems. By reasonably configuring the number of iterations, precision, and parallel threads, the system ensures that the solution to medium-difficulty problems can meet the optimization requirements while ensuring efficiency.For difficult problems, the complexity of the problem is high, and more computing resources and higher solution accuracy need to be allocated to ensure the accuracy and quality of the final solution. For example, the maximum number of iterations is set to 5000, and the number of iterations is increased to cope with the complexity of the problem. The convergence accuracy is set to 1e-6, and a higher convergence accuracy is required to ensure the accuracy of the solution. The relative error allowable range is set to 0.1%, and the accuracy requirement is strict to ensure the accuracy of the final result. The number of parallel processing threads is set to 8, and more parallel threads are used to speed up the solution. The heuristic algorithm is set to on, and the heuristic algorithm is enabled to improve the efficiency of solving complex problems. The initial slack variable is set to a smaller value to ensure that it starts from a more accurate initial condition and gradually converges to the optimal solution. By increasing the number of iterations, improving the accuracy, enabling multi-threaded processing and heuristic algorithms, the system can cope with difficult problems with high complexity and ensure that the quality of the solution is optimized within a reasonable time.

[0101] This application achieves rapid response to simple problems, balanced solutions to medium-difficulty problems, and efficient optimization of difficult problems by dynamically adjusting the solver parameters according to different difficulty levels and flexibly allocating computing resources and solution strategies, thereby improving the overall solution efficiency while ensuring the quality of the results.

[0102] In some optional implementations of this embodiment, the step of labeling the feature data with the difficulty level of the questions according to the preset conditions to obtain the labeled data set specifically includes:

[0103] analyzing the solution time of historical unit commitment problems in the characteristic data;

[0104] If the solution time is less than the first time length, the historical unit commitment problem is marked as a simple difficulty;

[0105] If the solution time is greater than or equal to the first time length and less than or equal to the second time length, the historical unit commitment problem is marked as medium difficulty;

[0106] If the solution time is greater than the second time length, marking the historical unit commitment problem as difficult;

[0107] The labeled feature data is used as the labeled data set.

[0108] In this embodiment, by analyzing the solution time of each unit combination problem in the historical data, as the main basis for judging its difficulty, the solution time reflects the complexity of the problem and the consumption of computing resources in the solution process. The longer the solution time, the greater the difficulty of the problem. Specifically, the first time length can be set to 500 seconds, and the second time length can be set to 1500 seconds. For unit combination problems with a solution time of less than 500 seconds, they are marked as "simple difficulty", that is, these problems can be solved in a shorter time, the complexity of the problem is low, and it is suitable to use lower resources and faster algorithms for solution; for problems with a solution time between 500 seconds and 1500 seconds, they are marked as "medium difficulty", these problems may involve more complex constraints or more variables, so the solution time is relatively long, but not too complex; for unit combination problems with a solution time of more than 1500 seconds, they are marked as "difficult difficulty", these problems are usually very complex, may involve multiple layers of constraints, high coupling between variables, etc., the solution process takes a long time, and the computing resource requirements are large. After completing the difficulty labeling, the feature dataset with the difficulty label is saved as a labeled dataset for subsequent machine learning model training. This dataset is used to train the question difficulty prediction model so that the system can automatically predict the difficulty level of new questions.

[0109] This application analyzes the solution time of historical unit combination problems and marks them as easy, medium or difficult. The system can build a dataset with difficulty labels to help the model learn how to predict the difficulty of problems based on feature data during training. This can dynamically adjust the solution strategy in the future unit combination problem solving, improve solution efficiency and optimize quality.

[0110] In some optional implementations of this embodiment, the step of constructing a question difficulty prediction model based on the labeled data set and a preset machine learning algorithm specifically includes:

[0111] Dividing the labeled data set into a training set and a test set;

[0112] According to the training set and the machine learning algorithm, an initial machine learning model is established and model training is performed to obtain a trained machine learning model;

[0113] The machine learning model is tested according to the test set, and parameters of the machine learning model are optimized according to the test results to obtain the problem difficulty prediction model.

[0114] In this embodiment, before building a machine learning model, the labeled data set needs to be divided into two parts: a training set and a test set. The training set is used to train the model, and the test set is used to evaluate the performance of the model. The common division ratio is 80% of the data set for training and 20% for testing; next, select a suitable machine learning algorithm, such as random forest, support vector machine (SVM) or neural network, to establish an initial machine learning model; then use the training set to train the model so that it can learn the relationship between different features and problem difficulty, and use the assigned test set to test the trained model to evaluate its performance on unseen data. By comparing the predicted difficulty level with the actual difficulty level, the performance indicators of the model such as accuracy, precision, and recall are calculated. According to the test results, the parameters of the model are adjusted and optimized.

[0115] This application divides the labeled data set into a training set and a test set, and uses machine learning algorithms for training and optimization, and finally builds a model that can accurately predict the difficulty of the unit combination problem. By continuously optimizing the model parameters, the prediction performance of the model is improved to ensure that it can provide reliable difficulty judgments in various problem scenarios, providing a basis for the subsequent dynamic adjustment of solver parameters.

[0116] In some optional implementations of this embodiment, the step of testing the machine learning model according to the test set, and optimizing the parameters of the machine learning model according to the test results to obtain the problem difficulty prediction model specifically includes:

[0117] Inputting the test set into the machine learning model to obtain an output result;

[0118] Verify and evaluate the output result according to the historical data to obtain the test result;

[0119] generating a model adjustment strategy based on the test results;

[0120] The parameters of the machine learning model are optimized according to the model adjustment strategy, and the machine learning model after the parameter optimization is used as the problem difficulty prediction model.

[0121] In this embodiment, after the model training is completed, the pre-divided test set is input into the trained machine learning model, and the model predicts the corresponding output result based on the input feature data, that is, the difficulty level of the unit combination problem; the predicted result output by the model is compared with the actual difficulty label in the test set (based on the actual difficulty level annotated with historical data), and the performance of the model is comprehensively evaluated by calculating indicators such as accuracy, precision, recall rate, F1 value, etc.; the pros and cons of the model are analyzed according to the test results, and the corresponding model adjustment strategy is generated. This strategy may include adjusting the hyperparameters of the model, modifying feature engineering, or selecting a new machine learning algorithm, etc., to improve the prediction ability of the model; according to the generated model adjustment strategy, the system adjusts or optimizes the hyperparameters of the machine learning model, such as increasing the number of decision trees, adjusting the learning rate, modifying the regularization parameter, etc., and the optimized model is trained again to become the final difficulty prediction model.

[0122] This application verifies the model through the test set, generates adjustment strategies based on the test results, and continuously optimizes the model parameters to finally obtain a difficulty prediction model with superior performance. Through this process, the system can ensure that the model has high accuracy and robustness when dealing with large-scale unit combination problems, thereby providing reliable support for the subsequent dynamic adjustment of solver parameters.

[0123] In some optional implementations of this embodiment, the step of extracting and preprocessing the historical data to obtain feature data specifically includes:

[0124] Extracting features from the historical data to obtain a feature vector set;

[0125] The feature vector set is subjected to missing value processing, outlier processing, data standardization and data dimension reduction to obtain the feature data.

[0126] In this embodiment, the key features related to solving the unit commitment problem are first extracted from the historical data, and converted into feature vectors that can be used for machine learning models. The features include the quotation, start-stop status, load demand, system constraints, etc. of the generator set. In actual data, some features may be missing. Missing value processing is an important step to ensure data integrity. Common processing methods include using the mean or median to fill missing values, or deleting records containing missing values. Outlier processing aims to identify and process abnormal points in the data. These abnormal points may be caused by input errors, system failures or extreme situations. Common processing methods include deleting abnormal values ​​or replacing them with values ​​within a reasonable range. The data ranges of different features may vary greatly. In order to avoid some features from having too much impact on the model due to large values, it is usually necessary to standardize the data. Common methods include normalizing or standardizing the feature data to zero mean unit variance. The purpose of data dimensionality reduction is to reduce the dimension of the feature vector, remove redundant or irrelevant features, and reduce the complexity of the model. Common dimensionality reduction techniques include principal component analysis (PCA), etc., which can extract the most representative features in the data.

[0127] Through feature extraction and preprocessing, this application systematically converts complex historical data into high-quality feature vectors that can be used in machine learning models. Through missing value processing, outlier processing, data standardization and dimensionality reduction, it ensures the integrity, consistency and simplicity of the data, providing a solid foundation for subsequent model training and prediction, thereby improving the performance and solution efficiency of the model.

[0128] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0129] The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, optimization technology for large-scale unit combination problems in large power systems, operation / interaction systems, mechatronics, etc. Artificial intelligence software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0130] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0131] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0132] Further references Figure 5 , as a response to the above Figure 2 The present application provides an embodiment of a device for optimizing large-scale unit combination problems in a power system. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0133] like Figure 5 As shown, the power system large-scale unit combination problem optimization device 300 described in this embodiment includes: an acquisition module 301, a processing module 302, a marking module 303, a construction module 304 and an implementation module 305. Among them:

[0134] An acquisition module 301 is used to acquire historical data of power system clearing cases;

[0135] The processing module 302 is used to extract and preprocess the historical data to obtain feature data;

[0136] The labeling module 303 is used to label the feature data with a difficulty level according to preset conditions to obtain a labeled data set;

[0137] A construction module 304 is used to construct a question difficulty prediction model based on the labeled data set and a preset machine learning algorithm;

[0138] The implementation module 305 is used to dynamically adjust the solver parameters of the large-scale unit commitment problem according to the problem difficulty prediction model.

[0139] The optimization device for large-scale unit combination problems in power systems provided in the present application not only improves the solution efficiency, but also significantly improves the optimization quality of large-scale unit combination problems, ensures the efficient use of system computing resources, reduces redundant calculations and resource waste, and thus optimizes the operating stability of the power system.

[0140] In some optional implementations of this embodiment, the construction module 304 is further used to:

[0141] Inputting the unit commitment problem to be predicted into the problem difficulty prediction model;

[0142] Outputting the problem difficulty level corresponding to the unit combination problem according to the problem difficulty prediction model;

[0143] The solver parameters are set according to the difficulty level of the problem.

[0144] The optimization device for large-scale unit combination problems in power systems provided in this application ensures that the system can reasonably allocate computing resources when solving large-scale unit combination problems, improve solution efficiency, and provide sufficient computing power to obtain accurate solutions for complex problems.

[0145] In some optional implementations of this embodiment, the difficulty level of the problem includes simple difficulty, medium difficulty and difficult difficulty, the solver parameters include the maximum number of iterations, convergence accuracy, relative error allowable range, number of parallel processing threads, heuristic algorithm switch and initial slack variable, and the implementation module 305 is further used to:

[0146] If the difficulty level of the problem is simple, the maximum number of iterations is set to the first number, the convergence accuracy is set to the first accuracy, the relative error allowable range is set to the first range, the number of parallel processing threads is set to the first value, the heuristic algorithm switch is set to off, and the initial slack variable is set to a larger value;

[0147] If the difficulty level of the problem is medium difficulty, the maximum number of iterations is set to the second number, the convergence accuracy is set to the second accuracy, the relative error allowable range is set to the second range, the number of parallel processing threads is set to the second value, the heuristic algorithm switch is set to on, and the initial slack variable is set to a medium value;

[0148] If the difficulty level of the problem is difficult, the maximum number of iterations is set to the third number, the convergence accuracy is set to the third accuracy, the relative error allowable range is set to the third range, the number of parallel processing threads is set to the third value, the heuristic algorithm switch is set to on, and the initial slack variable is set to a smaller value;

[0149] The second number is greater than the first number and less than the third number, the second precision is less than the first precision and greater than the third precision, the second range is less than the first range and greater than the third range, and the second value is greater than the first value and less than the third value.

[0150] The device for optimizing large-scale unit combination problems in power systems provided in the present application realizes rapid response to simple problems, balanced solution to medium-difficulty problems, and efficient optimization of difficult problems, thereby improving the overall solution efficiency while ensuring the quality of the results.

[0151] In some optional implementations of this embodiment, the construction module 304 is further used to:

[0152] analyzing the solution time of historical unit commitment problems in the characteristic data;

[0153] If the solution time is less than the first time length, the historical unit commitment problem is marked as a simple difficulty;

[0154] If the solution time is greater than or equal to the first time length and less than or equal to the second time length, the historical unit commitment problem is marked as medium difficulty;

[0155] If the solution time is greater than the second time length, marking the historical unit commitment problem as difficult;

[0156] The labeled feature data is used as the labeled data set.

[0157] The device for optimizing large-scale unit combination problems in power systems provided in this application can realize dynamic adjustment of solution strategies in solving future unit combination problems, thereby improving solution efficiency and optimization quality.

[0158] In some optional implementations of this embodiment, the marking module 303 is further used to:

[0159] Dividing the labeled data set into a training set and a test set;

[0160] According to the training set and the machine learning algorithm, an initial machine learning model is established and model training is performed to obtain a trained machine learning model;

[0161] The machine learning model is tested according to the test set, and parameters of the machine learning model are optimized according to the test results to obtain the problem difficulty prediction model.

[0162] The optimization device for large-scale unit combination problems in power systems provided in this application improves the prediction performance of the model by continuously optimizing model parameters, ensures that it can provide reliable difficulty judgments in various problem scenarios, and provides a basis for subsequent dynamic adjustment of solver parameters.

[0163] In some optional implementations of this embodiment, the construction module 304 is further used to:

[0164] Inputting the test set into the machine learning model to obtain an output result;

[0165] Verify and evaluate the output result according to the historical data to obtain the test result;

[0166] generating a model adjustment strategy based on the test results;

[0167] The parameters of the machine learning model are optimized according to the model adjustment strategy, and the machine learning model after the parameter optimization is used as the problem difficulty prediction model.

[0168] The optimization device for large-scale unit combination problems in power systems provided in this application ensures that the model has high accuracy and robustness when dealing with large-scale unit combination problems, thereby providing reliable support for subsequent dynamic adjustment of solver parameters.

[0169] In some optional implementations of this embodiment, the processing module 302 is further configured to:

[0170] Extracting features from the historical data to obtain a feature vector set;

[0171] The feature vector set is subjected to missing value processing, outlier processing, data standardization and data dimension reduction to obtain the feature data.

[0172] The optimization device for large-scale unit combination problems in power systems provided in this application ensures the integrity, consistency and simplicity of data through missing value processing, outlier processing, data standardization and dimensionality reduction, providing a solid foundation for subsequent model training and prediction, thereby improving the performance and solution efficiency of the model.

[0173] To solve the above technical problems, the present application also provides a computer device. Figure 6 , Figure 6 This is a basic structural block diagram of the computer device in this embodiment.

[0174] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 4 with components 41-43, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0175] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0176] The memory 41 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. equipped on the computer device 4. Of course, the memory 41 can also include both the internal storage unit of the computer device 4 and its external storage device. In this embodiment, the memory 41 is generally used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions of the optimization method for large-scale unit combination problems in power systems, etc. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or are to be output.

[0177] The processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other power system large-scale unit combination problem optimization chip in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run the computer-readable instructions or process data stored in the memory 41, such as computer-readable instructions for running the power system large-scale unit combination problem optimization method.

[0178] The network interface 43 may include a wireless network interface or a wired network interface. The network interface 43 is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0179] The computer equipment provided in this application obtains historical data of power system clearing cases, extracts features related to unit combination problems and performs preprocessing, and predicts the difficulty of the problem in combination with machine learning algorithms. While improving the efficiency of solving problems, it significantly improves the optimization quality of large-scale unit combination problems, ensures the efficient use of system computing resources, reduces redundant computing and resource waste, and thus optimizes the operational stability of the power system.

[0180] The present application also provides another embodiment, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the above-mentioned method for optimizing the large-scale unit combination problem of the power system.

[0181] The computer-readable storage medium provided by the present application obtains historical data of power system clearing cases, extracts features related to the unit combination problem and performs preprocessing, and combines machine learning algorithms to predict the difficulty of the problem. While improving the efficiency of solving, it significantly improves the optimization quality of large-scale unit combination problems, ensures the efficient use of system computing resources, reduces redundant computing and resource waste, and thus optimizes the operational stability of the power system.

[0182] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0183] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for optimizing large-scale unit commitment problems in a power system, characterized in that: The steps include: Obtain historical data on power system clearing cases; Extracting and preprocessing the historical data to obtain feature data; According to preset conditions, the feature data is labeled with a difficulty level to obtain a labeled data set; Constructing a problem difficulty prediction model based on the labeled data set and a preset machine learning algorithm; According to the problem difficulty prediction model, dynamic adjustment of solver parameters for large-scale unit commitment problems is achieved.

2. The method for optimizing large-scale unit commitment problem in a power system according to claim 1, characterized in that: The step of dynamically adjusting the solver parameters of the large-scale unit commitment problem according to the problem difficulty prediction model specifically includes: Inputting the unit commitment problem to be predicted into the problem difficulty prediction model; Outputting the problem difficulty level corresponding to the unit combination problem according to the problem difficulty prediction model; The solver parameters are set according to the difficulty level of the problem.

3. The method for optimizing large-scale unit commitment problem in a power system according to claim 2, characterized in that: The difficulty level of the problem includes simple difficulty, medium difficulty and difficult difficulty. The solver parameters include the maximum number of iterations, convergence accuracy, relative error tolerance, number of parallel processing threads, heuristic algorithm switch and initial slack variable. The step of setting the solver parameters according to the difficulty level of the problem specifically includes: If the difficulty level of the problem is simple, the maximum number of iterations is set to the first number, the convergence accuracy is set to the first accuracy, the relative error allowable range is set to the first range, the number of parallel processing threads is set to the first value, the heuristic algorithm switch is set to off, and the initial slack variable is set to a larger value; If the difficulty level of the problem is medium difficulty, the maximum number of iterations is set to the second number, the convergence accuracy is set to the second accuracy, the relative error allowable range is set to the second range, the number of parallel processing threads is set to the second value, the heuristic algorithm switch is set to on, and the initial slack variable is set to a medium value; If the difficulty level of the problem is difficult, the maximum number of iterations is set to the third number, the convergence accuracy is set to the third accuracy, the relative error allowable range is set to the third range, the number of parallel processing threads is set to the third value, the heuristic algorithm switch is set to on, and the initial slack variable is set to a smaller value; The second number is greater than the first number and less than the third number, the second precision is less than the first precision and greater than the third precision, the second range is less than the first range and greater than the third range, and the second value is greater than the first value and less than the third value.

4. The method for optimizing large-scale unit commitment problem in a power system according to claim 3, characterized in that: The step of labeling the feature data with a difficulty level according to the preset conditions to obtain a labeled data set specifically includes: analyzing the solution time of historical unit commitment problems in the characteristic data; If the solution time is less than the first time length, the historical unit commitment problem is marked as a simple difficulty; If the solution time is greater than or equal to the first time length and less than or equal to the second time length, the historical unit commitment problem is marked as medium difficulty; If the solution time is greater than the second time length, marking the historical unit commitment problem as difficult; The labeled feature data is used as the labeled data set.

5. The method for optimizing large-scale unit commitment problem in a power system according to claim 1, characterized in that: The step of constructing a question difficulty prediction model based on the labeled data set and the preset machine learning algorithm specifically includes: Dividing the labeled data set into a training set and a test set; According to the training set and the machine learning algorithm, an initial machine learning model is established and model training is performed to obtain a trained machine learning model; The machine learning model is tested according to the test set, and parameters of the machine learning model are optimized according to the test results to obtain the problem difficulty prediction model.

6. The method for optimizing large-scale unit commitment problem in a power system according to claim 5, characterized in that: The step of testing the machine learning model according to the test set, and optimizing the parameters of the machine learning model according to the test result to obtain the problem difficulty prediction model specifically includes: Inputting the test set into the machine learning model to obtain an output result; Verify and evaluate the output result according to the historical data to obtain the test result; generating a model adjustment strategy based on the test results; The parameters of the machine learning model are optimized according to the model adjustment strategy, and the machine learning model after the parameter optimization is used as the problem difficulty prediction model.

7. The method for optimizing large-scale unit commitment problem in a power system according to any one of claims 1 to 6, characterized in that: The step of extracting and preprocessing the historical data to obtain feature data specifically includes: Extracting features from the historical data to obtain a feature vector set; The feature vector set is subjected to missing value processing, outlier processing, data standardization and data dimension reduction to obtain the feature data.

8. An optimization device for large-scale unit combination problems in a power system, characterized in that: include: An acquisition module, used to obtain historical data of power system clearing cases; A processing module, used for performing feature extraction and preprocessing on the historical data to obtain feature data; A labeling module, used to label the feature data with a difficulty level according to preset conditions to obtain a labeled data set; A construction module, used to construct a question difficulty prediction model based on the labeled data set and a preset machine learning algorithm; The implementation module is used to dynamically adjust the solver parameters of the large-scale unit commitment problem according to the problem difficulty prediction model.

9. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the method for optimizing the large-scale unit combination problem of a power system as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the method for optimizing large-scale unit commitment problems in a power system according to any one of claims 1 to 7.

Images