Market clearing electric energy marginal cost prediction method and device based on big data analysis

Abstract

The invention discloses a market clearing electric energy marginal cost prediction method and device based on big data analysis, and the method comprises the steps: carrying out the preprocessing of historical system data, historical boundary data and historical weather data, and obtaining the preprocessing data; inputting the preprocessed data into a first integrated learning model, a first deep learning model and a first regression analysis model which are constructed in advance for training to obtain a second integrated learning model, a second deep learning model and a second regression analysis model, and performing fusion to obtain a marginal prediction model; and inputting the day-ahead prediction data into the marginal prediction model, and outputting a processing result of clearing the electric power spot market. According to the market clearing electric energy marginal cost prediction method based on big data analysis, large-scale complex data can be effectively processed and analyzed, and high-precision market cost prediction is provided.

Description

Technical Field

[0001] The present invention relates to the field of power market analysis and big data application, and in particular to a method and device for predicting the marginal cost of market-clearing electric energy based on big data analysis. Background Art

[0002] With the marketization of electricity and the widespread application of renewable energy, the complexity of electricity market operation is increasing. The prediction of the marginal cost of market-clearing electricity is of great significance for market participants to make scheduling decisions.

[0003] However, traditional forecasting methods often fail to fully utilize the available large-scale data resources, resulting in insufficient accuracy and reliability of forecast results. In addition, traditional methods are usually based on historical data, statistical models and expert experience for forecasting, but in the face of complex and changing power market environments and massive amounts of real-time data, these methods may not be able to fully explore and utilize the potential value in the data, resulting in limited accuracy and reliability of forecast results. Summary of the invention

[0004] The present invention aims to provide a method and device for predicting the marginal cost of market-clearing electricity based on big data analysis, so as to solve the technical problem of how to improve the accuracy of marginal cost prediction, thereby realizing real-time marginal cost prediction and providing decision support for market operation and power dispatching.

[0005] In order to solve the above technical problems, an embodiment of the present invention provides a method for predicting the marginal cost of market-clearing electric energy based on big data analysis, comprising:

[0006] Obtain historical system data, historical boundary data and historical weather data for the day-ahead market clearing of electricity spot;

[0007] Preprocessing the historical system data, the historical boundary data and the historical weather data by using missing value processing and data normalization methods to obtain preprocessed data;

[0008] The preprocessed data are respectively input into the pre-built first integrated learning model, the first deep learning model, and the first regression analysis model for training, so as to obtain the trained second integrated learning model, the second deep learning model, and the second regression analysis model in sequence;

[0009] fusing the second integrated learning model, the second deep learning model, and the second regression analysis model to obtain a marginal prediction model;

[0010] The day-ahead forecast data for clearing the electricity spot market is obtained, the day-ahead forecast data including system data, boundary data and meteorological data, and the data is input into the marginal forecast model for processing to obtain a first processing result, wherein the first processing result reflects the real-time marginal cost of clearing the electricity spot market.

[0011] As one preferred solution, the historical system data includes at least one or more of the following:

[0012] Power system load data, renewable energy power generation data, hydropower generation data, nuclear power generation data, inter-regional power transmission data, and market clearing data;

[0013] The historical boundary data includes at least one or more of the following:

[0014] Load forecast data, new energy power generation forecast data, and standby demand forecast data;

[0015] The historical weather data includes at least one or more of the following:

[0016] Temperature data, cloud cover data, wind data.

[0017] As one preferred solution, before the preprocessed data is respectively input into the pre-built first integrated learning model, the first deep learning model, and the first regression analysis model for training, the method further includes:

[0018] According to the characteristics of the preprocessed data, feature extraction is performed on the preprocessed data, wherein the extracted features at least include time features, weather features, energy type features, and historical market data features.

[0019] As one of the preferred schemes, the first integrated model includes a random forest model or an adaptive boosting algorithm, the first deep learning model includes a recurrent neural network and a long short-term memory network, the first regression analysis model includes a linear regression model, a ridge regression model or a lasso regression model, and the training of the first integrated learning model, the first deep learning model, and the first regression analysis model uses the Euclidean distance between the predicted value and the actual value as the loss function.

[0020] As one preferred solution, the preprocessed data is respectively input into a pre-built first integrated learning model, a first deep learning model, and a first regression analysis model for training, and further includes:

[0021] The performance of the first integrated learning model, the first deep learning model, and the first regression analysis model is evaluated, and the first integrated learning model, the first deep learning model, and the first regression analysis model are optimized according to the evaluation results, wherein the preset parameters include at least mean square error, root mean square error, mean absolute error, and determination coefficient.

[0022] As one preferred solution, after obtaining the first processing result, the method further includes planning power dispatching according to the first processing result:

[0023] If the first processing result meets the preset range, several new energy units in the power grid are controlled to adjust their own output values ​​to meet the needs of users, wherein the preset range includes three situations: a preset range selected by the user; a preset range obtained by the prediction law generated by the control center system based on historical data; a preset range obtained by weighted fusion based on the current power system's own scale, rated value, and output efficiency.

[0024] Another embodiment of the present invention provides a market-clearing electric energy marginal cost prediction device based on big data analysis, comprising:

[0025] An acquisition module is used to obtain historical system data, historical boundary data and historical weather data of the day-ahead market clearing of the power spot market;

[0026] A preprocessing module, used for preprocessing the historical system data, the historical boundary data and the historical weather data by using missing value processing and data normalization methods to obtain preprocessed data;

[0027] A training module, used to input the preprocessed data into the pre-built first integrated learning model, the first deep learning model, and the first regression analysis model for training, to obtain the trained second integrated learning model, the second deep learning model, and the second regression analysis model in turn;

[0028] A fusion module, used to fuse the second integrated learning model, the second deep learning model, and the second regression analysis model to obtain a marginal prediction model;

[0029] The output module is used to obtain the day-ahead forecast data for clearing the electricity spot market, the day-ahead forecast data including system data, boundary data and meteorological data, and input the data into the marginal forecast model for processing to obtain a first processing result, wherein the first processing result reflects the real-time marginal cost of clearing the electricity spot market.

[0030] As one preferred solution, the training module is further used for:

[0031] The performance of the first ensemble learning model, the first deep learning model, and the first regression analysis model is evaluated according to preset parameters, and the first ensemble learning model, the first deep learning model, and the first regression analysis model are optimized according to the evaluation results, wherein the preset parameters include at least mean square error, root mean square error, mean absolute error, and determination coefficient.

[0032] Another embodiment of the present invention provides a computer device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the market-clearing electricity marginal cost prediction method based on big data analysis as described above when executing the computer program.

[0033] Yet another embodiment of the present invention provides a computer-readable storage medium storing a computer program, wherein when the device where the computer-readable storage medium is located executes the computer program, the market-clearing electricity marginal cost prediction method based on big data analysis as described above is implemented.

[0034] Compared with the prior art, the embodiments of the present invention have the following advantages:

[0035] 1) The present invention can provide more accurate marginal cost prediction than traditional methods by comprehensively considering multiple factors and using advanced data analysis technology.

[0036] 2) The accurate marginal cost prediction of the present invention helps market participants to formulate more effective bidding and scheduling strategies, thereby improving the operating efficiency of the entire market.

[0037] 3) The present invention helps to better integrate renewable energy and support energy transformation and sustainable development goals by accurately predicting market conditions including large-scale new energy generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a flow chart of a method for predicting the marginal cost of market-clearing electric energy based on big data analysis in one embodiment of the present invention;

[0039] Figure 2 It is a structural schematic diagram of a market-clearing electric energy marginal cost prediction device based on big data analysis in one embodiment of the present invention;

[0040] Figure 3 It is a schematic diagram of a market-clearing electricity marginal cost prediction device based on big data analysis in one embodiment of the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] In the description of this application, the terms "first", "second", "third", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first", "second", "third", etc. may explicitly or implicitly include one or more of the feature. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0043] In the description of the present application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two elements. The terms "vertical", "horizontal", "left", "right", "upper", "lower" and similar expressions used herein are only for illustrative purposes, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0044] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as those commonly understood by those skilled in the art. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood by specific circumstances.

[0045] An embodiment of the present invention provides a method for predicting the marginal cost of market-clearing electricity based on big data analysis. For details, see Figure 1 , Figure 1 The flowchart of the method for predicting the marginal cost of market-clearing electric energy based on big data analysis provided in one embodiment of the present invention is shown. It includes steps S1 to S5:

[0046] S1. Obtain historical system data, historical boundary data and historical weather data for the day-ahead electricity spot market clearing;

[0047] In the embodiment of the present invention, preferably, the historical system data includes at least one or more of the following:

[0048] Power system load data, renewable energy power generation data, hydropower generation data, nuclear power generation data, inter-regional power transmission data, and market clearing data;

[0049] Historical boundary data includes at least one or more of the following:

[0050] Load forecast data, new energy power generation forecast data, and standby demand forecast data;

[0051] The historical weather data includes at least one or more of the following:

[0052] Temperature data, cloud cover data, wind data.

[0053] Among them, new energy power generation data include wind power and solar power generation power data, market clearing data include system clearing load and marginal cost related data, and new energy power generation forecast data include wind power and solar power generation power forecast data.

[0054] Specifically, the invention helps to better integrate renewable energy, support energy transformation and sustainable development goals by accurately predicting the market situation including a large amount of new energy power generation. Its high-precision prediction, intelligent integration, support for energy transformation and promotion of sustainable development will bring far-reaching impacts and positive changes to the energy industry, economic field and social field.

[0055] S2, preprocessing the historical system data, historical boundary data and historical weather data by using missing value processing and data normalization methods to obtain preprocessed data;

[0056] Specifically, missing value processing includes: filling missing data with linear interpolation or based on trend fitting methods;

[0057] Data normalization includes: To prevent the model from being biased during training, the features are normalized, such as minimum-maximum normalization or Z-score normalization.

[0058] In the embodiment of the present invention, preferably, before the preprocessed data is respectively input into the pre-built first integrated learning model, the first deep learning model, and the first regression analysis model for training, the method further includes:

[0059] According to the characteristics of the preprocessed data, feature extraction is performed on the preprocessed data, wherein the extracted features at least include time features, weather features, energy type features, and historical market data features.

[0060] Specifically, according to the characteristics of the data, key features are extracted, such as seasonal load changes, the impact of weather conditions, fuel price fluctuations, and changes in renewable energy generation.

[0061] Time features: Extract seasonal and trend features of time series data, such as hourly, daily, monthly, and seasonal indicators.

[0062] Weather characteristics: Combine meteorological data such as temperature, wind speed, and cloud coverage to analyze their impact on renewable energy power generation.

[0063] Energy type characteristics: Fuel price changes have a direct impact on marginal costs. Extract historical and real-time data on oil, coal and natural gas prices.

[0064] Historical market data: Using the load and marginal cost data in the historical clearing data, the historical feature sequence is constructed through the sliding window method.

[0065] S3, inputting the preprocessed data into the pre-built first integrated learning model, the first deep learning model, and the first regression analysis model for training, and obtaining the trained second integrated learning model, the second deep learning model, and the second regression analysis model in turn;

[0066] In an embodiment of the present invention, preferably, the first integrated model includes a random forest model or an adaptive boosting algorithm, the first deep learning model includes a recurrent neural network and a long short-term memory network, the first regression analysis model includes a linear regression model, a ridge regression model or a lasso regression model, and the training of the first integrated learning model, the first deep learning model, and the first regression analysis model uses the Euclidean distance between the predicted value and the actual value as the loss function.

[0067] Specifically, machine learning and statistical models are used to model and analyze feature data. Models include various types and are used to process different data.

[0068] Model 1: Ensemble learning method, using ensemble learning algorithms such as Random Forest and Adaptive Boost, because they can improve the accuracy and robustness of predictions when processing large data sets.

[0069] The random forest algorithm builds multiple decision trees and combines their prediction results to reduce the overfitting of a single model and improve the generalization ability of the model. Specifically, the algorithm is implemented through the following steps:

[0070] Sample sampling: Using the Bootstrap sampling method, multiple subsets are randomly selected from the original training set with replacement, and each subset is used to train a decision tree.

[0071] Feature sampling: While building each decision tree, for each split, only some features are randomly selected for split selection instead of considering all features.

[0072] Decision tree training: For each subset, build a decision tree.

[0073] Result fusion: For classification problems, the majority voting method is used to vote on the prediction results of all trees, and the category with the most votes is taken as the final prediction result; for regression problems, the average prediction of all trees is taken as the final prediction result.

[0074] The Adaboost (Adaptive Boosting) algorithm builds a strong classifier by combining multiple weak classifiers. Specifically, the Adaboost algorithm is implemented through the following steps: First, initialize the weight distribution of the training data. Assuming that there are N training sample data, each training sample is initially assigned the same weight. Then, multiple rounds of iterative training of weak classifiers are performed. For each round of iteration m (m = 1, 2, ..., M), perform the following steps: use a weighted data set to train the weak classifier, calculate the error rate of the weak classifier, calculate the weight coefficient of the weak classifier based on the error rate of the weak classifier, update the weight distribution of the training data set, and reduce the weight of correctly classified samples; for incorrectly classified samples, increase their weights. Finally, after M rounds of iterations, M weak classifiers and their corresponding weight coefficients are obtained. These weak classifiers are combined into a strong classifier in a weighted voting manner to obtain the final classifier.

[0075] Model 2: Deep learning methods, using recurrent neural networks (RNNs) and long short-term memory networks (LSTMs), which are particularly suitable for processing time series data and can capture long-term dependencies in the data.

[0076] RNN is a neural network specially designed for processing sequence data. It processes data in a serialized way to capture the temporal information and semantic information in the data. The network structure of RNN mainly consists of input layer, hidden layer and output layer. There are connections between the hidden layers of RNN, so that information can be transmitted between time steps. The training process of RNN includes two stages: forward propagation and back propagation: Forward propagation: Calculate the hidden state and output of each time step in chronological order. Back propagation: Update the parameters of the model through the gradient descent method.

[0077] LSTM solves the gradient vanishing and gradient exploding problems that are prone to occur in traditional RNNs during training by introducing a "gate" mechanism, thereby capturing long-term dependencies. The core of LSTM is a unit that contains a forget gate, an input gate, and an output gate. Each gate is a special structure used to control the flow of information.

[0078] Model 3: Regression analysis method, applying linear regression, ridge regression or Lasso regression models to predict marginal costs, especially when there is a linear relationship between features and targets.

[0079] The linear regression algorithm establishes a linear relationship between independent variables (features) and dependent variables (target variables) by finding the best fitting hyperplane (such as least squares method) in multidimensional space, and is used to predict the dependent variable values ​​corresponding to new or unknown independent variables.

[0080] The ridge regression or Lasso regression model is an improvement on the linear regression model, which introduces regularization to deal with biased estimates in collinear data analysis.

[0081] The input of the above model is the feature data obtained by extracting features from the preprocessed data, and the output is the predicted value of the marginal cost of clearing electricity in the electricity spot market at each time point corresponding to the predicted target time period.

[0082] In the embodiment of the present invention, preferably, the preprocessed data is respectively input into the pre-built first integrated learning model, the first deep learning model, and the first regression analysis model for training, and further includes:

[0083] The performance of the first integrated learning model, the first deep learning model, and the first regression analysis model is evaluated, and the first integrated learning model, the first deep learning model, and the first regression analysis model are optimized according to the evaluation results, wherein the preset parameters include at least mean square error, root mean square error, mean absolute error, and determination coefficient.

[0084] Furthermore, it also includes adjusting the structure of each model according to the verification results of the test set, such as increasing or decreasing the number of network layers, changing the activation function, adjusting the learning rate, etc.

[0085] S4, fusing the second integrated learning model, the second deep learning model, and the second regression analysis model to obtain a marginal prediction model;

[0086] Specifically, the present invention fuses the second integrated learning model, the second deep learning model, and the second regression analysis model through a simple average method, a weighted average method, or a stacking method to obtain a marginal cost prediction model. The simple average method is to perform a simple arithmetic average on the predicted values ​​of multiple models to obtain the final prediction result; the simple average method is simple and intuitive, and the calculation is simple. However, since it regards the predicted values ​​of all models as equally important, it may not be able to fully utilize the advantages of different models.

[0087] The weighted average method assigns different weights to the prediction values ​​of different models, and then averages the weighted prediction values ​​to obtain the final prediction result. The weight distribution can be determined based on the performance of the model, domain knowledge or other factors. The specific steps include:

[0088] 1) Run the integrated learning model, basic deep learning model and regression analysis model respectively to obtain their respective marginal cost prediction values;

[0089] 2) Assign a weight (e.g., 0.3, 0.5, 0.2) to each model’s prediction value based on its performance or other factors;

[0090] 3) Calculate the weighted average, that is, (prediction value 1*weight 1+prediction value 2*weight 2+prediction value 3*weight 3) / (weight 1+weight 2+weight 3), to get the final marginal cost prediction value.

[0091] The weighted average method can make full use of the advantages of different models and optimize the final prediction results by adjusting the weights.

[0092] The stacking method is also called meta-learning. It inputs the prediction results of multiple models into a meta-learner as new features, and the meta-learner gives the final prediction results. This method can make full use of the information of different models and obtain better prediction results through the optimization of the meta-learner. The specific steps include:

[0093] 1) Run the integrated learning model, basic deep learning model and regression analysis model respectively to obtain their respective marginal cost prediction values;

[0094] 2) Use the predicted values ​​of the three models as new features to construct a new dataset;

[0095] 3) Train a meta-learner (such as a linear regression model, neural network, etc.) on the new data set to obtain the final marginal cost prediction model.

[0096] The stacking method can make full use of the information of different models and obtain better prediction results through the optimization of the meta-learner.

[0097] Specifically, in the prior art, a model is usually used to train data of each dimension, while the data of multiple dimensions provided by the present invention means that more features can be used to describe an object or event. These features may have different information contents and correlations, thereby providing more information for making decisions.

[0098] Furthermore, different models may work based on different assumptions or algorithms. Some models may be good at capturing certain types of patterns, while other models may be good at capturing other types of patterns. By combining multiple models, the advantages of various models can be combined to improve the overall predictive ability. Using multiple models at the same time can reduce the reliance on a single model, thereby improving the robustness of the system. If one model performs poorly for some reason, other models may still be able to provide useful predictions.

[0099] S5. Obtain day-ahead forecast data for clearing the electricity spot market, the day-ahead forecast data including system data, boundary data and meteorological data, and input the data into the marginal forecast model for processing to obtain a first processing result, wherein the first processing result reflects the real-time marginal cost of clearing the electricity spot market.

[0100] In the embodiment of the present invention, preferably, after obtaining the first processing result, the method further includes planning power dispatching according to the first processing result:

[0101] If the first processing result meets the preset range, several new energy units in the power grid are controlled to adjust their own output values ​​to meet the needs of users, wherein the preset range includes three situations: a preset range selected by the user; a preset range obtained by the prediction law generated by the control center system based on historical data; a preset range obtained by weighted fusion based on the current power system's own scale, rated value, and output efficiency.

[0102] Specifically, based on the real-time marginal cost output by the model, the real-time supply and demand situation, price trends and other information of the electricity market are analyzed;

[0103] Formulate power dispatch plans based on real-time marginal costs and the actual situation of the power market. This may include adjusting the output of generators, optimizing the routing of power transmission, and exchanging power with neighboring power grids;

[0104] Communicate the formulated dispatch plan to relevant power operators and equipment operators and execute the dispatch plan.

[0105] Another embodiment of the present invention provides a market-clearing electric energy marginal cost prediction device based on big data analysis. For details, see Figure 2 , Figure 2The structure diagram of the market-clearing electric energy marginal cost prediction device based on big data analysis provided in one embodiment of the present invention is shown, which includes the following modules: an acquisition module 11, a preprocessing module 12, a training module 13, a fusion module 14, and an output module 15, wherein:

[0106] An acquisition module 11 is used to acquire historical system data, historical boundary data and historical weather data of the day-ahead market clearing of the power spot market;

[0107] A preprocessing module 12 is used to preprocess the historical system data, historical boundary data and historical weather data by using missing value processing and data normalization methods to obtain preprocessed data;

[0108] The training module 13 is used to input the preprocessed data into the pre-built first integrated learning model, the first basic deep learning model, and the first basic regression analysis model for training, and obtain the trained second integrated learning model, the second basic deep learning model, and the second basic regression analysis model in turn;

[0109] A fusion module 14 is used to fuse the second integrated learning model, the second basic deep learning model, and the second basic regression analysis model to obtain a marginal cost prediction model;

[0110] The output module 15 is used to obtain the day-ahead forecast data for clearing the electricity spot market, wherein the day-ahead forecast data includes system data, boundary data and meteorological data, and is input into the marginal forecast model for processing to obtain a first processing result, wherein the first processing result reflects the real-time marginal cost of clearing the electricity spot market.

[0111] In the embodiment of the present invention, preferably, the training module is further used for:

[0112] The performance of the first ensemble learning model, the first deep learning model, and the first regression analysis model is evaluated according to preset parameters, and the first ensemble learning model, the first deep learning model, and the first regression analysis model are optimized according to the evaluation results, wherein the preset parameters include at least mean square error, root mean square error, mean absolute error, and determination coefficient.

[0113] Another embodiment of the present invention provides a market clearing electric energy marginal cost prediction device based on big data analysis. For details, please refer to Figure 3 , Figure 3A schematic diagram of a market-clearing electricity marginal cost prediction device based on big data analysis provided in one embodiment of the present invention is shown, which includes a processor 21, a memory 22, and a computer program stored in the memory 22 and configured to be executed by the processor 21. When the processor 21 executes the computer program, the steps in the above-mentioned market-clearing electricity marginal cost prediction method based on big data analysis are implemented, for example Figure 1 or, when the processor 21 executes the computer program, the functions of the modules in the above-mentioned device embodiments are implemented, such as the acquisition module 11.

[0114] Exemplarily, the computer program may be divided into one or more modules, which are stored in the memory 22 and executed by the processor 21 to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program in the market-clearing electric energy marginal cost prediction device based on big data analysis.

[0115] The processor 21 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor 21 is the control center of the device, and uses various interfaces and lines to connect various parts of the market-clearing electric energy marginal cost prediction device based on big data analysis.

[0116] The memory 22 can be used to store the computer program and / or module, and the processor 21 realizes various functions of the market-clearing electric energy marginal cost prediction device based on big data analysis by running or executing the computer program and / or module stored in the memory 22, and calling the data stored in the memory 22. The memory 22 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory 22 can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0117] Wherein, if the module integrated by the market-clearing electric energy marginal cost prediction device based on big data analysis is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0118] A person 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 a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0119] Accordingly, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to perform the steps in the market-clearing electricity marginal cost prediction method based on big data analysis as described in the above embodiment, for example Figure 1 Steps S1 to S5 described in .

[0120] Compared with the prior art, the embodiments of the present invention have the following advantages:

[0121] 1) The present invention can provide more accurate marginal cost prediction than traditional methods by comprehensively considering multiple factors and using advanced data analysis technology.

[0122] 2) The accurate marginal cost prediction of the present invention helps market participants to formulate more effective bidding and scheduling strategies, thereby improving the operating efficiency of the entire market.

[0123] 3) The present invention helps to better integrate renewable energy and support energy transformation and sustainable development goals by accurately predicting market conditions including large-scale new energy generation.

[0124] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A method for predicting the marginal cost of market-clearing electricity based on big data analysis. It is characterized in that The following steps are involved: Obtain historical system data, historical boundary data and historical weather data for the day-ahead market clearing of electricity spot; Preprocessing the historical system data, the historical boundary data and the historical weather data by using missing value processing and data normalization methods to obtain preprocessed data; The preprocessed data are respectively input into the pre-built first integrated learning model, the first deep learning model, and the first regression analysis model for training, so as to obtain the trained second integrated learning model, the second deep learning model, and the second regression analysis model in sequence; fusing the second integrated learning model, the second deep learning model, and the second regression analysis model to obtain a marginal prediction model; The day-ahead forecast data for clearing the electricity spot market is obtained, the day-ahead forecast data including system data, boundary data and meteorological data, and the data is input into the marginal forecast model for processing to obtain a first processing result, wherein the first processing result reflects the real-time marginal cost of clearing the electricity spot market.

2. According to the method for predicting the marginal cost of market-clearing electric energy based on big data analysis according to claim 1, It is characterized in that The historical system data includes at least one or more of the following: Power system load data, renewable energy power generation data, hydropower generation data, nuclear power generation data, inter-regional power transmission data, and market clearing data; The historical boundary data includes at least one or more of the following: Load forecast data, new energy power generation forecast data, and standby demand forecast data; The historical weather data includes at least one or more of the following: Temperature data, cloud cover data, wind data.

3. According to the method for predicting the marginal cost of market-clearing electric energy based on big data analysis according to claim 1, It is characterized in that Before the preprocessed data are respectively input into the pre-built first integrated learning model, the first deep learning model, and the first regression analysis model for training, the method further includes: According to the characteristics of the preprocessed data, feature extraction is performed on the preprocessed data, wherein the extracted features at least include time features, weather features, energy type features, and historical market data features.

4. The method for predicting the marginal cost of market-clearing electric energy based on big data analysis according to claim 1, It is characterized in that The first integrated model includes a random forest model or an adaptive boosting algorithm, the first deep learning model includes a recurrent neural network and a long short-term memory network, the first regression analysis model includes a linear regression model, a ridge regression model or a lasso regression model, and the training of the first integrated learning model, the first deep learning model and the first regression analysis model uses the Euclidean distance between the predicted value and the actual value as the loss function.

5. The method for predicting the marginal cost of market-clearing electric energy based on big data analysis according to claim 1, It is characterized in that The step of inputting the preprocessed data into the pre-built first integrated learning model, the first deep learning model, and the first regression analysis model for training also includes: The performance of the first integrated learning model, the first deep learning model, and the first regression analysis model is evaluated, and the first integrated learning model, the first deep learning model, and the first regression analysis model are optimized according to the evaluation results, wherein the preset parameters include at least mean square error, root mean square error, mean absolute error, and determination coefficient.

6. The method for predicting the marginal cost of market-clearing electric energy based on big data analysis according to claim 1, It is characterized in that After obtaining the first processing result, the method further includes planning power dispatching according to the first processing result: If the first processing result meets the preset range, several new energy units in the power grid are controlled to adjust their own output values ​​to meet the needs of users, wherein the preset range includes three situations: a preset range selected by the user; a preset range obtained by the prediction law generated by the control center system based on historical data; a preset range obtained by weighted fusion based on the current power system's own scale, rated value, and output efficiency.

7. A device for predicting the marginal cost of market-clearing electric energy based on big data analysis, It is characterized in that include: An acquisition module is used to obtain historical system data, historical boundary data and historical weather data of the day-ahead market clearing of the power spot market; A preprocessing module, used for preprocessing the historical system data, the historical boundary data and the historical weather data by using missing value processing and data normalization methods to obtain preprocessed data; A training module, used to input the preprocessed data into the pre-built first integrated learning model, the first deep learning model, and the first regression analysis model for training, to obtain the trained second integrated learning model, the second deep learning model, and the second regression analysis model in turn; A fusion module, used to fuse the second ensemble learning model, the second deep learning model, and the second regression analysis model by a simple averaging method, a weighted averaging method, or a stacking method to obtain a marginal prediction model; The output module is used to obtain the day-ahead forecast data for clearing the electricity spot market, the day-ahead forecast data including system data, boundary data and meteorological data, and input the data into the marginal forecast model for processing to obtain a first processing result, wherein the first processing result reflects the real-time marginal cost of clearing the electricity spot market.

8. The market-clearing electric energy marginal cost prediction device based on big data analysis according to claim 7, It is characterized in that The training module is also used for: The performance of the first ensemble learning model, the first deep learning model, and the first regression analysis model is evaluated according to preset parameters, and the first ensemble learning model, the first deep learning model, and the first regression analysis model are optimized according to the evaluation results, wherein the preset parameters include at least mean square error, root mean square error, mean absolute error, and determination coefficient.

9. A market-clearing electricity marginal cost prediction device based on big data analysis, It is characterized in that It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the market-clearing electricity marginal cost prediction method based on big data analysis as described in any one of claims 1 to 6.

10. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, it implements the market-clearing electricity marginal cost prediction method based on big data analysis as described in any one of claims 1 to 6.

Images