Method for evaluating influence of decision-making behavior of electricity market participant on market price
By constructing a decision-making behavior evaluation model for power market participants, combining multiple linear regression and random forest algorithms, the problem of insufficient evaluation accuracy and adaptability in the existing technology is solved, high accuracy and real-timeness are achieved, and intuitive visualization is provided. Show to support the healthy development of the power market.
Patent Information
- Application Number
- CN202410928472.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-11
- Publication Date
- 2024-11-08
AI Technical Summary
When evaluating the impact of decision-making behaviors of power market participants on market prices, the prior art lacks accuracy, lacks dynamicity and interactivity, makes it difficult to process big data, and lacks visual presentation tools, resulting in poor accuracy and adaptability of model predictions.
By acquiring and preprocessing the decision-making behavior and market price data of power market participants, an evaluation model is constructed, and a combination of multivariate linear regression and random forest algorithm is used to perform feature selection and dataset division, optimize model parameters and features, and use cross-validation and Model integration technology to realize real-time data updates and visual presentation.
It improves the accuracy and real-time evaluation, enhances the adaptability and visualization of the model, and provides more accurate and intuitive information support for power market decisions.
Smart Images

Figure CN118917904A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of power market analysis and prediction, and in particular to a method for evaluating the impact of decision-making behaviors of power market participants on market prices. Background Art
[0002] As the power market continues to develop and become more complex, the impact of power market participants' decision-making behavior on market prices has become increasingly significant. However, existing evaluation methods often fail to accurately capture the decision-making behavior of power market participants and their impact on market prices. Therefore, a more accurate and comprehensive method is needed to evaluate the impact of power market participants' decision-making behavior on market prices.
[0003] In the field of evaluating the impact of decision-making behavior of electricity market participants on market prices, existing technologies mainly rely on traditional econometric models and statistical analysis methods. These technologies usually collect historical data of the electricity market, such as transaction data, market quotations, supply and demand data, etc., and then use regression analysis, time series analysis and other methods to predict and evaluate market prices. In addition, some studies have also used some simple machine learning algorithms, such as linear regression and logistic regression, to simulate the decision-making behavior of electricity market participants.
[0004] The main problems or deficiencies:
[0005] Insufficient model accuracy: Traditional econometric models and statistical analysis methods often assume that the electricity market is static or follows a certain distribution. However, in the real world, the electricity market is complex and changeable, and the decision-making behavior of participants is also affected by many factors, which limits the prediction accuracy of traditional methods.
[0006] Lack of dynamism and interactivity: Existing technologies rarely take into account the dynamic interactions and mutual influences between power market participants. The power market is a typical network market, where the decision-making behaviors of participants are interrelated and influenced, and traditional methods often ignore this interactivity.
[0007] Limited ability to process big data: As the power market continues to develop, the amount of data generated is also increasing. Traditional technologies often have difficulty processing and analyzing large amounts of data, especially real-time data, which limits their application in power market analysis and forecasting.
[0008] Poor model generalization ability: Although some technologies that use machine learning algorithms have improved the accuracy of predictions to a certain extent, due to the complexity and variability of the electricity market, these models have poor generalization ability and are difficult to adapt to new market environments and conditions.
[0009] Lack of visual display and intuitive understanding: Existing technologies rarely provide intuitive visual display tools, making it difficult for users to intuitively understand the impact of the decision-making behavior of electricity market participants on market prices, which is not conducive to decision makers making accurate judgments.
[0010] In summary, existing technologies have certain limitations and shortcomings in assessing the impact of decision-making behaviors of electricity market participants on market prices, and more advanced and accurate methods need to be developed to solve these problems. Summary of the invention
[0011] In order to solve the above technical problems, the present invention provides a method for evaluating the impact of decision-making behavior of power market participants on market prices, comprising:
[0012] Obtain decision-making behavior data of each participant in the electricity market and market price data;
[0013] Perform data preprocessing on the acquired data;
[0014] Construct an evaluation model of the impact of power market participants’ decision-making behavior on market prices;
[0015] The preprocessed data is evaluated based on the evaluation model.
[0016] Furthermore, the decision-making behavior data includes:
[0017] Bidding data, including bidding time, bidding quantity, bidding price and other information;
[0018] Quotation data, including quotation time, quotation quantity, quotation price and other information;
[0019] Bargaining data, including information on the bargaining process and bargaining results;
[0020] Other relevant decision-making behavior data, such as transaction frequency, transaction volume, etc.;
[0021] The market price data include:
[0022] Real-time electricity price data: reflects the current market electricity price level;
[0023] Historical electricity price data: used to analyze price trends and cyclical characteristics;
[0024] Market supply and demand data: including information such as electricity supply and demand balance and electricity load changes.
[0025] Furthermore, the data acquisition method includes using data acquisition tools to obtain data in real time from the transaction system, quotation system, and monitoring system of the power market.
[0026] Furthermore, the data preprocessing includes data cleaning, data sorting, data classification and data standardization; wherein;
[0027] Data cleaning includes deduplication, processing of missing values, and removal of noise and outliers; data organization includes unifying the data format and unit and sorting the data; data classification includes classification according to time, participants and other business needs; data standardization includes minimum-maximum standardization, Z-score standardization and decimal calibration standardization to standardize the data.
[0028] Furthermore, the establishment of the evaluation model includes the following steps:
[0029] (1) Feature Selection and Engineering
[0030] Through correlation analysis and feature importance assessment methods, features with significant impact on market prices are selected from the original data, mainly the decision-making behavior characteristics of power market participants, which specifically include the power generation plan of power generation companies, the power purchase strategy of power sales companies, and the power consumption behavior of power users, which will have an impact on market prices. Features may include historical transaction data, predicted transaction data, and risk preferences of market participants.
[0031] Further processing and transformation of the selected features to improve the performance of the model, including feature scaling, feature encoding, feature combination, and feature dimensionality reduction;
[0032] (2) Network model selection
[0033] Based on the decision-making behavior characteristics of the above-mentioned electricity market participants, the multivariate linear regression in the regression analysis model and the random forest algorithm in the machine learning model are specifically selected and combined to construct an evaluation model for electricity market price forecasting.
[0034] (3) Model training
[0035] The data set is divided into training set, validation set and test set; the training set is used to train the model, the validation set is used to adjust the model parameters and select the best model, and the test set is used to evaluate the performance of the model; initial values are set for the parameters in the model; the model is trained using the training set data, and the model parameters are updated through an iterative optimization algorithm to minimize the error between the model's predicted results and the actual results.
[0036] Furthermore, it also includes evaluating and optimizing the model, wherein the evaluation process includes data set division, evaluation index selection and cross-validation; specifically,
[0037] The data set is divided into training set, validation set and test set in a ratio of 70%:15%:15%;
[0038] Evaluation indicators are determined according to the type of problem such as classification, regression, clustering and the characteristics of the data set, including accuracy, precision, recall, F1 value, and area under the ROC curve AUC; cross-validation is used to divide the data set into multiple subsets, and the model is trained multiple times to evaluate the performance of the model.
[0039] Furthermore, it also includes model optimization, including feature selection, parameter tuning, model integration and data enhancement;
[0040] Use filtering method, wrapping method and embedded method to screen the features of the model;
[0041] Grid search and random search methods are used to tune parameters;
[0042] Use Bagging and Boosting model integration methods to combine and integrate the prediction results of multiple models to reduce the variance or bias of the model;
[0043] The training data set of the model is expanded by rotating, flipping, cropping, and adding noise, so that the model can learn more data distribution and features.
[0044] The method for evaluating the impact of decision-making behavior of power market participants on market prices of the present invention has the following effects and advantages:
[0045] 1. High accuracy: This patent can accurately simulate the decision-making process of power market participants and accurately evaluate the impact of these decision-making behaviors on market prices by building a decision-making behavior model of power market participants and combining big data analysis and artificial intelligence algorithms. This high accuracy provides strong data support for the stable operation of the power market and the formulation of power policies.
[0046] 2. Strong real-time performance: This patent establishes a real-time data update mechanism to ensure that the evaluation method can timely reflect the latest changes in the power market. This means that no matter what kind of fluctuations occur in the power market, this patent can respond quickly and provide decision makers with the latest market information and forecast results.
[0047] 3. Comprehensiveness: By integrating power market data from different sources, such as power grid data, meteorological data, economic data, etc., this patent can provide a more comprehensive understanding of the operation of the power market and the decision-making behavior of participants. This comprehensiveness not only improves the accuracy of the assessment, but also provides decision makers with richer information references.
[0048] 4. Good visualization effect: This patent uses visualization tools to display the evaluation results, allowing users to intuitively understand the impact of the decision-making behavior of power market participants on market prices. This intuitiveness not only facilitates decision makers to make accurate judgments, but also improves the readability and comprehensibility of the evaluation results.
[0049] 5. Flexibility and adaptability: The evaluation method of this patent has good flexibility and adaptability. By adjusting the model parameters and algorithm settings, it can adapt to different market environments and conditions and meet various evaluation needs. In addition, with the continuous advancement of technology and changes in the market environment, this patent can also be iteratively updated and optimized to maintain its advancement and effectiveness.
[0050] 6. High decision-making support value: The evaluation results provided by this patent are of great reference value to decision makers in the power market. By understanding the impact of the decision-making behavior of power market participants on market prices, decision makers can formulate more reasonable and effective power policies and market strategies to promote the healthy development of the power market. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] Figure 1 It is a schematic diagram of the process of the present invention;
[0052] Figure 2 The model output is shown in the form of a ROC curve. DETAILED DESCRIPTION
[0053] The principles and features of the present invention are described below in conjunction with the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.
[0054] refer to Figure 1-2 The present invention provides a method for evaluating the impact of decision-making behavior of power market participants on market prices, comprising:
[0055] S1: Obtain the decision-making behavior data and market price data of each participant in the electricity market;
[0056] Use dedicated data collection tools to collect real-time or regular data on participants' decision-making behaviors, such as bidding, quoting, bargaining, etc., from the trading system, quotation system, and monitoring system of the power market. At the same time, collect market price data, including real-time electricity prices, transaction volumes, market supply and demand, etc.
[0057] 1) Data collection objectives
[0058] The main goal of data collection is to obtain the decision-making behavior data of each participant in the electricity market and market price data in order to provide a basis for subsequent evaluation and analysis.
[0059] 2) Data collection scope
[0060] The scope of data collection includes but is not limited to the following aspects:
[0061] Participants’ decision-making behavior data:
[0062] Bidding data: including bidding time, bidding electricity, bidding price and other information.
[0063] Quotation data: including quotation time, quotation electricity, quotation price and other information.
[0064] Bargaining data: including information such as the bargaining process and bargaining results.
[0065] Other relevant decision-making behavior data, such as transaction frequency, transaction volume, etc.
[0066] Market price data:
[0067] Real-time electricity price data: reflects the current market electricity price level.
[0068] Historical electricity price data: used to analyze price trends and cyclical characteristics.
[0069] Market supply and demand data: including information such as electricity supply and demand balance and electricity load changes.
[0070] 3) Data collection methods, including real-time collection, periodic collection and collection from multiple data sources:
[0071] Real-time collection: Use data collection tools to obtain real-time data from the trading system, quotation system, and monitoring system of the power market. Collection tools include data collection and analysis systems, power data monitoring and collection instruments, web crawlers, API tools, database management tools, and professional data collection software;
[0072] Among them, data acquisition and analysis systems: This type of system usually uses data acquisition technology to obtain data from the trading system, quotation system, and monitoring system of the power market in real time or regularly. The data is processed and analyzed through data analysis tools to obtain the true situation of market operation.
[0073] Such systems can automatically collect, integrate and clean data, improving the efficiency and accuracy of data collection.
[0074] Among them, power data monitoring and acquisition instruments are specially used for the collection and monitoring of power data, such as the Ankerui brand pollution source online monitoring data acquisition and transmission instrument (AF-HK100-4G), etc. These acquisition instruments can monitor the power production, transmission and consumption in real time and collect corresponding quantitative data.
[0075] It is usually installed in power generation plants, transmission lines, substations and other equipment to provide accurate power data.
[0076] Among them, web crawlers: For publicly accessible power market data, web crawlers can be used to automatically obtain it. Web crawlers simulate browser behavior, visit relevant websites of the power market and extract data on the pages. Such tools are very effective for large-scale data capture, such as obtaining market price data from the official website of the power market.
[0077] Among them, API tools: Many power markets or related institutions provide APIs (application programming interfaces) for accessing and acquiring data. API tools can quickly collect data by calling these interfaces, which is a relatively simple and efficient way to acquire data.
[0078] Among them, database management tools: For the power market data already stored in the database, the database management tools can be used to import and export the data.
[0079] This type of tool is not only used to manage databases, but also has functions such as data cleaning, integration and analysis, which helps in the subsequent processing of data.
[0080] Among them, professional data collection software: For enterprise-level users, professional data collection software can be used for data collection and integration. This type of software is usually comprehensive and supports the collection and integration of multiple data sources, such as 147 collection software. They can provide a full range of data collection solutions to meet complex data collection needs.
[0081] The data collected in real time is highly timely and accurate, and can reflect market trends in a timely manner.
[0082] Periodic collection: For some data that is not updated in real time, such as historical electricity price data, market supply and demand data, etc., a fixed collection cycle (such as daily, weekly, monthly, etc.) can be set for regular collection. Regularly collected data helps analyze market trends and cyclical characteristics.
[0083] Collection from multiple data sources: Considering the comprehensiveness and accuracy of the data, data can be collected from multiple data sources, such as power trading centers, power regulatory agencies, power companies, etc.
[0084] Data from different data sources can complement and verify each other, improving data integrity and reliability.
[0085] 4) Data collection process
[0086] Determine the scope and objectives of data collection. Select appropriate data collection tools and methods. Set data collection parameters and cycles. Connect to data sources and start data collection procedures. Receive and store data in real time or periodically. Clean, organize, classify and standardize the collected data.
[0087] S2: preprocess the acquired data;
[0088] Clean the collected data to remove noise and outliers. Classify and standardize the data to ensure consistency and comparability.
[0089] In the data preprocessing process of evaluating the impact of decision-making behavior of power market participants on market prices, in order to ensure the quality, consistency and comparability of data, a series of data cleaning, sorting, classification and standardization processing are required. The data preprocessing includes data cleaning, data sorting, data classification and data standardization;
[0090] The following is the detailed process of data preprocessing:
[0091] Among them, data cleaning specifically includes, removing duplicate data: check whether there are duplicate records in the data set. If so, choose to keep unique records or merge them according to business needs. Handling missing values: check whether there are missing values (null values or NULL values) in the data set, and select appropriate filling methods according to the number and distribution of missing values, such as mean filling, median filling, mode filling or interpolation. Removing noise and outliers: Identify and process noise and outliers in the data, which may be caused by equipment failure, human error or data transmission error. Outliers can be detected and removed by setting thresholds, using statistical methods (such as Z-score, IQR method, etc.) or machine learning algorithms (such as isolation forest, etc.).
[0092] The data collation process includes: Data format unification: Ensure that the format of all data in the data set is consistent, such as date and time format, number format, etc. For inconsistent formats, conversion and unification are required. Data unit unification: Check whether the units of each field in the data set are unified. If not, unit conversion is required to ensure the comparability of the data in subsequent analysis. Data sorting: Sort the data set according to business needs for better subsequent analysis and processing.
[0093] The data classification process includes, 1) Time classification: According to the timestamp or date field, the data is classified in chronological order, such as by day, week, month, year, etc. This helps to analyze the time trend of market prices. 2) Participant classification: According to the type of electricity market participants (such as generators, electricity buyers, retailers, etc.), the data is classified by participant type. This helps to analyze the impact of different participants' decision-making behaviors on market prices. 3) Other classifications: According to business needs, data can also be classified according to other fields (such as region, voltage level, etc.) for more detailed analysis.
[0094] The data standardization process includes: 1) Minimum-maximum standardization: Map the data to [0,1] through linear transformation. This method is suitable for situations where the distribution range of the original data is not large. 2) Z-score standardization: Convert the data to a standard normal distribution, that is, a distribution with a mean of 0 and a standard deviation of 1. This method is suitable for situations where the distribution range of the original data is large or skewed. 3) Decimal calibration standardization: Standardize by moving the decimal point of the data, moving the decimal point of the data by k places so that the absolute value of the data is between [1,10).
[0095] Through the above data preprocessing process, the quality, consistency and comparability of the decision-making behavior data and market price data of electricity market participants can be ensured, providing strong support for subsequent data analysis and evaluation.
[0096] S3: Construct an evaluation model of the impact of decision-making behavior of electricity market participants on market prices;
[0097] Combining economic principles, statistical methods and machine learning algorithms, an evaluation model for the impact of decision-making behavior of electricity market participants on market prices is constructed. The model considers multiple dimensions of participants' decision-making behavior, such as quotation strategy, transaction frequency, transaction volume, etc., as well as external factors such as market structure and supply and demand relationship. The model parameters are adjusted and optimized through training data to improve the model's prediction accuracy and generalization ability. The establishment of the model specifically includes the following processes,
[0098] 1) Feature Selection and Engineering
[0099] Feature selection: Select features that have a significant impact on market prices from the original data. For example, the decision-making behavior of power market participants, including the power generation plan of power generation companies, the power purchasing strategy of power sales companies, and the power consumption behavior of power users, will have an impact on market prices. Features may include historical transaction data of market participants, predicted transaction data, risk preferences of market participants, etc. This can be achieved through methods such as correlation analysis and feature importance assessment.
[0100] Feature Engineering: Further processing and transformation of selected features to improve the performance of the model. This may include feature scaling, feature encoding (such as one-hot encoding, label encoding, etc.), feature combination, feature dimensionality reduction (such as PCA, t-SNE, etc.), etc.
[0101] 2) Model selection
[0102] According to the characteristics of the problem and the characteristics of the data, we select the appropriate model, specifically the multivariate linear regression in the regression analysis model and the random forest in the machine learning model, and use them in combination to build a more powerful and accurate evaluation model for electricity market price prediction.
[0103] Multiple linear regression: Use multiple independent variables (such as supply and demand, power generation costs, policy factors, etc.) to predict electricity market prices. By collecting and analyzing the historical data of these independent variables, a regression equation is established to predict the changing trend of market prices.
[0104] Random forest: Random forest is an ensemble learning method based on decision trees. It improves the accuracy and stability of predictions by building multiple decision trees and combining their prediction results. In power market forecasting, random forest can be used to process multiple independent variables and predict market prices.
[0105] Before building a statistical model, feature selection is usually required to identify features that have a significant impact on market prices. This can be achieved through methods such as correlation analysis, mutual information, and decision trees. For example, the correlation coefficient or mutual information value between each feature and the market price can be calculated, and features that are highly correlated with the market price can be selected as inputs to the model.
[0106] After selecting a statistical model, you need to use historical data to train and verify the model. By comparing the difference between the model's prediction results and the actual market price (such as mean square error, accuracy, etc.), you can evaluate the performance of the model. Based on the evaluation results, you can adjust the model parameters or choose another more suitable model.
[0107] In addition, methods such as cross-validation and grid search can be used to optimize the model's parameter settings to improve the model's prediction accuracy and generalization ability.
[0108] It should be noted that the statistical models and methods mentioned above are not isolated, they can be used in combination with each other to build a more powerful and accurate electricity market price forecasting model. At the same time, due to the complexity and uncertainty of the electricity market, no model can fully and accurately predict future market prices. Therefore, in practical applications, it is necessary to comprehensively consider multiple factors and flexibly use various models and methods to make predictions and decisions.
[0109] 3) Model training
[0110] Data partitioning: Divide the dataset into training set, validation set and test set. The training set is used to train the model, the validation set is used to adjust the model parameters and select the best model, and the test set is used to evaluate the performance of the model.
[0111] Parameter initialization: Set initial values for parameters in the model.
[0112] Model training: Use the training set data to train the model, and update the model parameters through iterative optimization algorithms (such as gradient descent, stochastic gradient descent, Adam, etc.) to minimize the error between the model's predicted results and the actual results.
[0113] S4: Evaluate and optimize the preprocessed data based on the evaluation model.
[0114] Use cross-validation and other methods to evaluate the model and verify its accuracy and reliability. Iterate and optimize the model based on the evaluation results to improve the model's predictive performance and robustness.
[0115] Model evaluation and optimization:
[0116] 1. Model evaluation
[0117] Model evaluation mainly measures the performance of the model through a series of indicators to determine whether the model has achieved the expected effect.
[0118] 1). Dataset division
[0119] Before evaluating the model, the dataset needs to be divided into training set, validation set, and test set. The ratio can be 70%:15%:15%.
[0120] 2) Choose the evaluation metric
[0121] The choice of evaluation indicators should be determined according to the type of problem (such as classification, regression, clustering, etc.) and the characteristics of the data set. Commonly used evaluation indicators include accuracy, precision, recall, F1 value, area under the ROC curve (AUC), etc.
[0122] 3) Cross-validation
[0123] Cross-validation is a commonly used model evaluation method that evaluates the performance of a model by dividing the dataset into multiple subsets and training the model multiple times. Common cross-validation methods include K-fold cross-validation and leave-one-out cross-validation.
[0124] 2. Model Optimization
[0125] Model optimization mainly improves the performance of the model by adjusting model parameters, changing model structure or using other techniques.
[0126] 1) Feature selection
[0127] Feature selection is one of the important steps to optimize model performance. By selecting the most representative features, the complexity of the model can be reduced and the generalization ability of the model can be improved. Commonly used feature selection methods include filtering methods, wrapping methods, and embedded methods.
[0128] 2) Parameter Tuning
[0129] Machine learning models usually contain multiple parameters, and the selection of these parameters has a significant impact on the performance of the model. Parameter tuning is the process of finding the optimal parameter combination by adjusting model parameters. Common parameter tuning methods include grid search, random search, etc.
[0130] 3) Model Integration
[0131] Model integration is to improve the performance of the model by combining the prediction results of multiple models. Commonly used model integration methods include bagging, boosting, etc. By integrating multiple models, the variance or bias of the model can be reduced, thereby improving the stability and generalization ability of the model.
[0132] 4) Data Augmentation
[0133] Data augmentation is a method to improve model performance by increasing the amount and diversity of training data. Common data augmentation methods include rotation, flipping, cropping, adding noise, etc. Through data augmentation, the training data set of the model can be expanded, so that the model can learn more data distribution and features, thereby improving the robustness and generalization ability of the model.
[0134] S5: Evaluation result output:
[0135] The evaluation results are output in the form of charts, reports, etc., showing the specific impact and trend of the participants' decision-making behavior on market prices. The evaluation results can be provided to power market managers, regulatory agencies, etc. as a reference for decision-making.
[0136] Input layer: Receives pre-processed participant decision behavior data and market price data.
[0137] Feature extraction layer: extracts and transforms the input data to extract features that have a significant impact on market prices.
[0138] Model calculation layer: Based on the extracted features, a machine learning algorithm is used to calculate the impact of participants’ decision-making behavior on market prices.
[0139] Output layer: Output the calculation results in the form of charts, reports, etc.
[0140] Evaluation result output:
[0141] When performing model evaluation, the output of the results is usually presented in the form of charts to more intuitively understand and explain the performance of the model. The following are some common evaluation result output charts and their detailed descriptions:
[0142] 1. Confusion Matrix
[0143] Description: The confusion matrix is used for classification problems and shows the comparison between the number of instances of each class predicted by the model and the number of instances of the actual class.
[0144] Graph: A two-dimensional table that usually includes four parts: True Positives (TP), False Positives (FP), True Negatives (TN) and False Negatives (FN).
[0145]
[0146] 2.ROC curve (Receiver Operating Characteristic Curve)
[0147] Description: ROC curve is used to evaluate the performance of classification models, especially when the performance of the model at different thresholds is of interest.
[0148] Graph: A two-dimensional graph with the horizontal axis representing the false positive rate (FPR) and the vertical axis representing the true positive rate (TPR), also known as the recall rate.
[0149] A two-dimensional coordinate system where the curve runs from the lower left corner (FPR=0, TPR=0) to the upper left corner (FPR=1, TPR=1).
[0150] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for evaluating the impact of decision-making behavior of power market participants on market prices, characterized in that: include, Obtain decision-making behavior data of each participant in the electricity market and market price data; Perform data preprocessing on the acquired data; Construct an evaluation model of the impact of power market participants’ decision-making behavior on market prices; The preprocessed data is evaluated based on the evaluation model.
2. A method for evaluating the impact of decision-making behavior of power market participants on market prices as claimed in claim 1, characterized in that: The decision-making behavior data includes bidding data, quotation data, quotation data and other relevant decision-making behavior data; the bidding data includes bidding time, bidding electricity, and bidding price information; Quotation data, including quotation time, quotation power, and quotation price information; Quotation data, including information on the bargaining process and bargaining results; other relevant decision-making behavior data, including transaction frequency and transaction volume; The market price data includes real-time electricity price data, historical electricity price data and historical electricity price data; the real-time electricity price data is used to reflect the current market electricity price level; the historical electricity price data is used to analyze price change trends and cyclical characteristics; the historical electricity price data includes electricity supply and demand balance and electricity load change information.
3. A method for evaluating the impact of decision-making behavior of power market participants on market prices as claimed in claim 1, characterized in that: Data acquisition methods include using data collection tools to obtain real-time data from the trading system, quotation system, and monitoring system of the power market.
4. A method for evaluating the impact of decision-making behavior of power market participants on market prices as claimed in claim 1, characterized in that: The data preprocessing includes data cleaning, data sorting, data classification and data standardization; wherein, Data cleaning includes deduplication, processing of missing values, and removal of noise and outliers; data organization includes unifying the data format and unit and sorting the data; data classification includes classification according to time, participants and other business needs; data standardization includes minimum-maximum standardization, Z-score standardization and decimal calibration standardization to standardize the data.
5. A method for evaluating the impact of decision-making behavior of power market participants on market prices as claimed in claim 1, characterized in that: The establishment of the evaluation model includes the following steps: (1) Feature selection and engineering: Select features that have a significant impact on market prices from raw data through correlation analysis and feature importance assessment methods, including historical trading data, predicted trading data, and risk preferences of market participants; further process and transform the selected features, including feature scaling, feature encoding, feature combination, and feature dimensionality reduction; (2) Network model selection: Based on the above-mentioned decision-making behavior characteristics of electricity market participants, the multivariate linear regression and random forest algorithms are combined to construct an evaluation model for electricity market price forecasting.
6. A method for evaluating the impact of decision-making behavior of power market participants on market prices as claimed in claim 1, characterized in that: It also includes evaluating and optimizing the model, wherein the evaluation process includes data set division, evaluation indicator selection and cross-validation; specifically, The data set is divided into training set, validation set and test set in a ratio of 70%:15%:15%; Evaluation indicators are determined according to the type of problem such as classification, regression, clustering and the characteristics of the data set, including accuracy, precision, recall, F1 value, and area under the ROC curve AUC; cross-validation is used to divide the data set into multiple subsets, and the model is trained multiple times to evaluate the performance of the model.
7. A method for evaluating the impact of decision-making behavior of power market participants on market prices as claimed in claim 1, characterized in that: It also includes optimizing the model. Including feature selection, parameter tuning, model integration and data enhancement; Use filtering method, wrapping method and embedded method to screen the features of the model; Grid search and random search methods are used to tune parameters; Use Bagging and Boosting model integration methods to combine and integrate the prediction results of multiple models to reduce the variance or bias of the model; The training data set of the model is expanded by rotating, flipping, cropping, and adding noise, so that the model can learn more data distribution and features.
