Power price prediction method based on dynamic integrated self-adaption and related equipment

Abstract

The invention relates to the technical field of time sequence prediction, integrated deep learning and electricity market analysis, and discloses an electricity price prediction method and related equipment based on dynamic integrated self-adaptation, and the method comprises the steps: obtaining multivariate electricity price data, and carrying out the preprocessing of the multivariate electricity price data, and obtaining a training data set; constructing a dynamic integrated adaptive electricity price prediction model, training a plurality of base learner models by using a training data set, evaluating the performance of the plurality of base learner models by using a verification set, dynamically adjusting the weight of a model combination, and obtaining a final weight; inputting the test data into the plurality of base learner models, and combining the final weight to obtain a final prediction value; according to the method, the nonlinear statistical characteristics in the time sequence can be effectively captured by constructing the dynamic integrated adaptive price prediction model, the dynamic nature and non-stationarity of data can be better coped with, and accurate prediction of the short-term power price is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of time series prediction, integrated deep learning and power market analysis, and specifically to a method for predicting power prices based on dynamic integrated adaptation and related equipment. Background Art

[0002] Electricity price forecasting is an important task in the electricity market and is of great significance to power producers, traders, consumers and regulators. Accurate electricity price forecasting can help market participants make effective plans and decisions, optimize resource allocation, reduce operating costs, and improve economic benefits.

[0003] At present, the short-term electricity price forecasting methods mainly use forecasting methods such as neural networks. However, since electricity prices are affected by many factors, such as weather conditions, supply and demand, holidays, macroeconomic indicators, etc., these factors make electricity prices highly volatile and non-stationary, and changes in these factors may cause changes in market price behavior, thereby reducing forecast accuracy and efficiency. At the same time, traditional short-term electricity price forecasting methods have limited data processing capabilities, fail to capture long-term dependencies in data, and lack the ability to handle the dynamics, non-stationarity and high volatility of electricity prices, resulting in low accuracy in electricity price forecasting. Using neural network methods to address such problems has not proposed an effective solution. Summary of the invention

[0004] In order to overcome the defects of the above-mentioned prior art, the purpose of the present invention is to provide a method and related equipment for electricity price prediction based on dynamic integrated adaptation, so as to solve the technical problem of how to improve the accuracy of short-term electricity price prediction in the prior art.

[0005] The present invention is achieved through the following technical solutions: In a first aspect, the present invention provides a method for predicting electricity prices based on dynamic integrated self-adaptation, comprising: Acquire multivariate electricity price data, and preprocess the multivariate electricity price data to obtain a training data set; Build a dynamic integrated adaptive electricity price prediction model, use the training data set to train multiple base learner models, use the validation set to evaluate the performance of multiple base learner models, dynamically adjust the weight of the model combination, and obtain the final weight; The test data is input into multiple base learner models and combined with the final weights to obtain the final prediction value.

[0006] Preferably, in the step of obtaining multivariate electricity price data and preprocessing the multivariate electricity price data to obtain a training data set, the specific process of preprocessing is as follows: The quartile method is used to detect and remove abnormal data in the multivariate electricity price data. The missing values ​​in the multivariate electricity price data are filled using the regional mean filling method. The Z-score standardization is used to adjust the numerical range to the [-1,1] interval to complete the preprocessing of the multivariate electricity price data.

[0007] Preferably, in the step of constructing a dynamic integrated adaptive electricity price prediction model, using a training data set to train multiple base learner models, using a validation set to evaluate the performance of multiple base learner models, dynamically adjusting the weight of the model combination, and obtaining the final weight, the specific process is as follows: Taking multivariate electricity price data as input variables, multiple base learner models are trained through key features of the input variables; The prediction error is calculated using the validation data obtained by evaluating the performance of multiple base learner models using the validation set, and the weight of the model combination is dynamically adjusted according to the prediction error to obtain the final weight.

[0008] Furthermore, the multiple base learner models include a long short-term memory network model LSTM, a gated recurrent unit GRU and a convolutional neural network CNN, wherein each base learner model is individually trained according to the training data and continuously updates its internal parameters during the training process to optimize the prediction performance of time series data.

[0009] Furthermore, the validation data obtained by evaluating the performance of multiple base learner models using the validation set calculates the prediction error, and the weight of the model combination is dynamically adjusted according to the prediction error. In the step of obtaining the final weight, the specific process is as follows: The prediction error is set to the past prediction error at each time step, and the combination weight is determined based on the past prediction error at each time step and a genetic factor. The expression is as follows:

[0010] In the formula, represents the forgetting factor; v represents the length of the sliding window used to calculate the prediction error; represents the prediction error of the mth base model at time step t; The weight adjustment is calculated based on the average error and cumulative error of all base learner models, and its expression is as follows:

[0011] In the formula, represents the cumulative prediction error of the mth base model at time step t; is the average cumulative prediction error of the mth model up to time step t; Based on the iteration weights and weight adjustment amount of the previous step, the iteration weights of multiple base learner models are calculated, and their expressions are as follows:

[0012] In the formula, represents the weight of the mth base model at time step k+1; represents the weight adjustment of the mth base model at time step k; Represents the learning rate, which is used to control the update speed of the weight adjustment amount; The iterative weights of multiple base learner models are input into the exponential weighted average algorithm to obtain the smoothed weights, which are expressed as follows:

[0013] In the formula, Indicates how fast the weight decreases. The smaller the value, the faster the weight decreases. is the EMA value at time k+1, that is, the final weight at time k+1; is the recursive weight at time k+1; Represents the EMA value at time k; Normalize the smoothing weight to get the final weight, which is expressed as follows: .

[0014] Preferably, in the step of inputting the test data into multiple base learner models and combining the final weights to obtain the final prediction value, the final prediction value expression is as follows:

[0015] In the formula, represents the final weight; Represents the prediction results within multiple base learner models.

[0016] In a second aspect, the present invention further provides a power price prediction system based on dynamic integrated self-adaptation, comprising: A data preprocessing module is used to obtain multivariate electricity price data and preprocess the multivariate electricity price data to obtain a training data set; The model processing module is used to build a dynamic integrated adaptive electricity price prediction model, train multiple base learner models using the training data set, evaluate the performance of multiple base learner models using the validation set, dynamically adjust the weight of the model combination, and obtain the final weight; The model test output module is used to input the test data into multiple base learner models and combine the final weights to obtain the final prediction value.

[0017] Preferably, the model processing module includes a training module and a verification module; A training module, used to take multivariate electricity price data as input variables and train multiple base learner models through key features of the input variables; The verification module is used to calculate the prediction error using the verification data obtained by evaluating the performance of multiple base learner models using the verification set, and dynamically adjust the weight of the model combination according to the prediction error to obtain the final weight.

[0018] In a third aspect, the present invention also provides a mobile terminal comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the dynamic integrated adaptive electricity price prediction method as described above when executing the computer program.

[0019] In a fourth aspect, the present invention further provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method for predicting electricity price based on dynamic integrated adaptation as described above.

[0020] Compared with the prior art, the present invention has the following beneficial technical effects: The present invention provides a method for predicting electricity prices based on dynamic integration and adaptation. By constructing an effective network of multiple base models, the complementary effect between the prediction results of different base models is maximized, and the combined weight of each base model is calculated using past prediction errors and forgetting factors. These weights are adjusted according to the historical performance of each model to better reflect their current prediction capabilities. The final weight is obtained using smoothing factors and normalization. The prediction results of each individual learner are fused using a weighted average method to obtain the final prediction output. The present invention can effectively capture the nonlinear statistical characteristics in the time series by constructing a dynamic integrated adaptive price prediction model, better cope with the dynamic and non-stationary nature of the data, and achieve accurate prediction of short-term electricity prices.

[0021] Furthermore, by dynamically adjusting the weight of each individual learner, more weight is allocated to the model with better performance according to the past prediction errors. This can better cope with the dynamic and non-stationary nature of the data and improve the prediction accuracy. The present invention has high computational efficiency in terms of runtime performance and can determine the combined weights without complex optimization processes. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flow chart of a method for predicting electricity prices based on dynamic integrated self-adaptation in an embodiment of the present invention; Figure 2 It is a framework diagram of the power price prediction method based on dynamic integrated self-adaptation in an embodiment of the present invention; Figure 3 It is a table diagram of time series data after outlier processing in an embodiment of the present invention; Figure 4A flowchart of training and verification of a dynamically integrated adaptive price prediction model in an embodiment of the present invention; Figure 5 A framework diagram for dynamically integrating adaptive price prediction model training and verification in an embodiment of the present invention; Figure 6 A schematic diagram of electricity price prediction using a dynamically integrated adaptive price prediction model in an embodiment of the present invention; Figure 7 This is a structural diagram of a power price prediction system based on dynamic integrated self-adaptation in an embodiment of the present invention; Figure 8 This is a schematic diagram of the structure of the model processing module in an embodiment of the present invention; In the figure: 1. Data preprocessing module; 2. Model processing module; 3. Model test output module; 21. Training module; 22. Verification module. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings 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. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0024] The purpose of the present invention is to provide a method and related equipment for predicting electricity prices based on dynamic integrated self-adaptation, so as to solve the technical problem of how to improve the accuracy of short-term electricity price prediction in the prior art.

[0025] The present invention is further described in detail below in conjunction with the accompanying drawings: Example 1 See also Figure 1 and Figure 2 In one embodiment of the present invention, a method for predicting electricity prices based on dynamic integrated self-adaptation is provided, comprising: Step 1, obtaining multivariate electricity price data, and preprocessing the multivariate electricity price data to obtain a training data set; This example uses the electricity price data of the EPEX French market as the simulation source. Considering the high volatility of electricity prices and the frequent price spikes, in order to ensure the stability of the data, the data is preprocessed to remove the influence of the mean and volatility of the data.

[0026] Specifically, the specific process of preprocessing is as follows: The quartile method was used to detect and remove abnormal data in the multivariate electricity price data. The missing values ​​in the multivariate electricity price data were filled using the regional mean filling method. The Z-score standardization was used to adjust the value range to the [-1,1] interval, and 39432 data were obtained. This covers electricity price data under the influence of multiple factors, providing data support for subsequent model training. Some data in the electricity price data set are as follows Figure 3 shown.

[0027] Step 2: construct a dynamic integrated adaptive electricity price prediction model, use the training data set to train multiple base learner models, use the validation set to evaluate the performance of multiple base learner models, dynamically adjust the weight of the model combination, and obtain the final weight; This embodiment establishes multiple independent neural network models to enhance the diversity of individual learners and obtain multiple electricity price forecast values; defines a forgetting factor β (0<β<1) to control the influence of past prediction errors. At the same time, the prediction error of each individual learner in the past period of time is recorded. The combined weight of each individual learner is calculated using the past prediction error and the forgetting factor. Subsequently, a smoothing factor is used to mitigate the violent fluctuations in the weight update process to obtain the final dynamic weight; the prediction results of each individual learner are integrated through the final dynamic weight to generate the final prediction result.

[0028] Specifically, according to Figure 4 and Figure 5 As shown, the specific process is as follows: S1, taking multivariate electricity price data as input variables, and training multiple base learner models through the key features of the input variables; Specifically, the multiple base learner models include a long short-term memory network model LSTM, a gated recurrent unit GRU and a convolutional neural network CNN, wherein each base learner model is trained individually according to the training data, and its internal parameters are continuously updated during the training process to optimize the prediction performance of time series data.

[0029] S2, uses the validation data obtained by evaluating the performance of multiple base learner models using the validation set to calculate the prediction error, and dynamically adjusts the weight of the model combination according to the prediction error to obtain the final weight.

[0030] Specifically, the specific process is as follows: The prediction error is set to the past prediction error at each time step, and the combination weight is determined based on the past prediction error at each time step and a genetic factor. The expression is as follows:

[0031] In the formula, represents the forgetting factor, and in this embodiment, β is set to 0.75; v represents the length of the sliding window used to calculate the prediction error; represents the prediction error of the mth base model at time step t; The weight adjustment is calculated based on the average error and cumulative error of all base learner models, and its expression is as follows:

[0032] In the formula, represents the cumulative prediction error of the mth base model at time step t; is the average cumulative prediction error of the mth model up to time step t; Based on the iteration weights and weight adjustment amount of the previous step, the iteration weights of multiple base learner models are calculated, and their expressions are as follows:

[0033] In the formula, represents the weight of the mth base model at time step k+1; represents the weight adjustment of the mth base model at time step k; Represents the learning rate, which is used to control the update speed of the weight adjustment amount. In this embodiment, α is set to 0.3; The iterative weights of multiple base learner models are input into the exponential weighted average algorithm to obtain the smoothed weights, which are expressed as follows:

[0034] In the formula, Indicates how fast the weight decreases. The smaller the value, the faster the weight decreases. In this embodiment, μ is 0.93; is the EMA value at time k+1, that is, the final weight at time k+1; is the recursive weight at time k+1; Represents the EMA value at time k; Normalize the smoothing weight to get the final weight, which is expressed as follows: .

[0035] Step 3: Input the test data into multiple base learner models and combine the final weights to obtain the final prediction value.

[0036] Specifically, the final predicted value expression is as follows:

[0037] In the formula, represents the final weight; Represents the prediction results within multiple base learner models.

[0038] In this embodiment, model training and reasoning are performed. The power price is predicted based on the power price prediction method based on dynamic integrated self-adaptation provided in this embodiment. The data comes from the day-ahead power market. The time period is from January 1, 2012 to June 30, 2016. The sequence data from January 1, 2012 to December 31, 2015 (a total of 1,461 days) is used for training and verification, and the sequence data from January 1, 2016 to June 30, 2016 (a total of 182 days) is used for testing. In the experiment, the parameters α and V are set to 0.85 and 4 respectively, and the initial weight is set to 0. According to Figure 6 As shown in the figure, it is the prediction result of the electricity price 24 hours before the prediction day. Compared with the actual electricity price, it can be seen that the electricity price predicted by this model can accurately reflect the price change trend of the day and fits well with the actual value.

[0039] The analysis of the day-ahead electricity price forecast results in this embodiment is shown in Table 1. The prediction results using the DE-EPF method, LSTM method, GRU method and Transformer method are tested respectively. It can be found that the prediction performance of the DE-EPF method in the present invention is significantly better than other comparison methods. Compared with other models, the average absolute error is reduced by at least 28%, the root mean square error is reduced by at least 23%, and the average absolute percentage error is reduced by at least 14%.

[0040] Table 1. Experimental results of model comparison on day-ahead electricity price dataset

[0041] In summary, the method for predicting electricity prices based on dynamic integrated adaptation provided in this embodiment constructs an effective network of multiple base models to maximize the complementary effects between the prediction results of different base models, and uses past prediction errors and forgetting factors to calculate the combined weights of each base model. These weights are adjusted according to the historical performance of each model to better reflect their current prediction capabilities. The final weights are obtained using smoothing factors and normalization. The prediction results of each individual learner are fused using a weighted average to obtain the final prediction output. The present invention can effectively capture the nonlinear statistical characteristics in time series by constructing a dynamic integrated adaptive price prediction model, better cope with the dynamics and non-stationarity of data, and achieve accurate prediction of short-term electricity prices.

[0042] The dynamic ensemble learning method in this embodiment is an effective means to help neural networks overcome the high volatility and non-stationarity problems in electricity price forecasting. Through dynamic ensemble learning, multiple different types of base learners can be combined to adapt to changing market conditions and improve the accuracy and robustness of the overall forecast. The present invention mainly studies the short-term electricity price forecasting method DE-EPF based on dynamic ensemble adaptation to achieve accurate prediction of multivariate data of electricity prices.

[0043] Example 2 according to Figure 7 As shown, the present invention also provides a power price prediction system based on dynamic integrated self-adaptation, comprising: The data preprocessing module 1 is used to obtain multivariate electricity price data and preprocess the multivariate electricity price data to obtain a training data set; Model processing module 2 is used to build a dynamic integrated adaptive electricity price prediction model, train multiple base learner models using a training data set, evaluate the performance of multiple base learner models using a validation set, dynamically adjust the weight of the model combination, and obtain the final weight; The model test output module 3 is used to input the test data into multiple base learner models and combine the final weights to obtain the final prediction value.

[0044] Specifically, according to Figure 8 As shown, the model processing module 2 includes a training module 21 and a verification module 22; A training module 21 is used to use multivariate electricity price data as input variables and train multiple base learner models through key features of the input variables; The verification module 22 is used to calculate the prediction error using the verification data obtained by evaluating the performance of multiple base learner models using the verification set, and dynamically adjust the weight of the model combination according to the prediction error to obtain the final weight.

[0045] Example 3 The present invention also provides a mobile terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, such as a power price prediction program based on dynamic integrated adaptation.

[0046] When the processor executes the computer program, the steps of the above-mentioned method for predicting electricity prices based on dynamic integrated adaptation are implemented, for example: Acquire multivariate electricity price data, and preprocess the multivariate electricity price data to obtain a training data set; Build a dynamic integrated adaptive electricity price prediction model, use the training data set to train multiple base learner models, use the validation set to evaluate the performance of multiple base learner models, dynamically adjust the weights of the model combination, and obtain the final weights.

[0047] The test data is input into multiple base learner models and combined with the final weights to obtain the final prediction value.

[0048] Alternatively, when the processor executes the computer program, the functions of each module in the above system are realized, for example: The data preprocessing module 1 is used to obtain multivariate electricity price data and preprocess the multivariate electricity price data to obtain a training data set; Model processing module 2 is used to build a dynamic integrated adaptive electricity price prediction model, train multiple base learner models using a training data set, evaluate the performance of multiple base learner models using a validation set, dynamically adjust the weight of the model combination, and obtain the final weight; The model test output module 3 is used to input the test data into multiple base learner models and combine the final weights to obtain the final prediction value.

[0049] Wherein, the model processing module 2 includes a training module 21 and a verification module 22; A training module 21 is used to use multivariate electricity price data as input variables and train multiple base learner models through key features of the input variables; The verification module 22 is used to calculate the prediction error using the verification data obtained by evaluating the performance of multiple base learner models using the verification set, and dynamically adjust the weight of the model combination according to the prediction error to obtain the final weight.

[0050] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program in the mobile terminal.

[0051] For example, the computer program may be divided into a data preprocessing module 1, a model processing module 2, and a model test output module 3; The specific functions of each module are as follows: The data preprocessing module 1 is used to obtain multivariate electricity price data and preprocess the multivariate electricity price data to obtain a training data set; Model processing module 2 is used to build a dynamic integrated adaptive electricity price prediction model, train multiple base learner models using a training data set, evaluate the performance of multiple base learner models using a validation set, dynamically adjust the weight of the model combination, and obtain the final weight; The model test output module 3 is used to input the test data into multiple base learner models and combine the final weights to obtain the final prediction value.

[0052] Wherein, the model processing module 2 includes a training module 21 and a verification module 22; A training module 21 is used to use multivariate electricity price data as input variables and train multiple base learner models through key features of the input variables; The verification module 22 is used to calculate the prediction error using the verification data obtained by evaluating the performance of multiple base learner models using the verification set, and dynamically adjust the weight of the model combination according to the prediction error to obtain the final weight.

[0053] The mobile terminal may be a computing device such as a desktop computer, a notebook, a palm computer, a cloud server, etc. The mobile terminal may include, but is not limited to, a processor and a memory.

[0054] The processor 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. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the mobile terminal, and uses various interfaces and lines to connect various parts of the entire mobile terminal.

[0055] The memory may be used to store the computer programs and / or modules, and the processor implements various functions of the mobile terminal by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory.

[0056] The memory may mainly include a program storage area and a data storage area, wherein the program storage area may 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 may store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmartMediaCard, SMC), a secure digital (SecureDigital, SD) card, a flash card (FlashCard), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0057] Example 4 The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for predicting electricity prices based on dynamic integrated adaptation are implemented.

[0058] If the module / unit integrated in the mobile terminal 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.

[0059] Based on this understanding, the present invention implements all or part of the processes in the above method, and can also be completed by instructing related hardware through a computer program, and the computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the above-mentioned aggregate reinforcement learning resource scheduling method can be implemented. Among them, 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.

[0060] The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0061] It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media does not include electrical carrier signals and telecommunication signals.

[0062] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting electricity prices based on dynamic integrated self-adaptation, characterized in that: include: Acquire multivariate electricity price data, and preprocess the multivariate electricity price data to obtain a training data set; Build a dynamic integrated adaptive electricity price prediction model, use the training data set to train multiple base learner models, use the validation set to evaluate the performance of multiple base learner models, dynamically adjust the weight of the model combination, and obtain the final weight; The test data is input into multiple base learner models and combined with the final weights to obtain the final prediction value.

2. The method for predicting electricity prices based on dynamic integrated self-adaptation according to claim 1, characterized in that: In the step of obtaining multivariate electricity price data and preprocessing the multivariate electricity price data to obtain a training data set, the specific process of preprocessing is as follows: The quartile method is used to detect and remove abnormal data in the multivariate electricity price data. The missing values ​​in the multivariate electricity price data are filled using the regional mean filling method. The Z-score standardization is used to adjust the numerical range to the [-1,1] interval to complete the preprocessing of the multivariate electricity price data.

3. The method for predicting electricity prices based on dynamic integrated self-adaptation according to claim 1, characterized in that: In the step of constructing a dynamic integrated adaptive electricity price prediction model, using a training data set to train multiple base learner models, using a validation set to evaluate the performance of multiple base learner models, dynamically adjusting the weight of the model combination, and obtaining the final weight, the specific process is as follows: Taking multivariate electricity price data as input variables, multiple base learner models are trained through key features of the input variables; The prediction error is calculated using the validation data obtained by evaluating the performance of multiple base learner models using the validation set, and the weight of the model combination is dynamically adjusted according to the prediction error to obtain the final weight.

4. The method for predicting electricity prices based on dynamic integrated self-adaptation according to claim 3, characterized in that: The multiple base learner models include a long short-term memory network model LSTM, a gated recurrent unit GRU and a convolutional neural network CNN, wherein each base learner model is trained individually according to training data and its internal parameters are continuously updated during the training process to optimize the prediction performance of time series data.

5. The method for predicting electricity prices based on dynamic integrated self-adaptation according to claim 3, characterized in that: The validation data obtained by using the validation set to evaluate the performance of multiple base learner models calculates the prediction error, and dynamically adjusts the weight of the model combination according to the prediction error to obtain the final weight. The specific process is as follows: The prediction error is set to the past prediction error at each time step, and the combination weight is determined based on the past prediction error at each time step and a genetic factor. The expression is as follows: In the formula, represents the forgetting factor; v represents the length of the sliding window used to calculate the prediction error; represents the prediction error of the mth base model at time step t; The weight adjustment is calculated based on the average error and cumulative error of all base learner models, and its expression is as follows: In the formula, represents the cumulative prediction error of the mth base model at time step t; is the average cumulative prediction error of the mth model up to time step t; Based on the iteration weights and weight adjustment amount of the previous step, the iteration weights of multiple base learner models are calculated, and their expressions are as follows: In the formula, represents the weight of the mth base model at time step k+1; represents the weight adjustment of the mth base model at time step k; Represents the learning rate, which is used to control the update speed of the weight adjustment amount; The iterative weights of multiple base learner models are input into the exponential weighted average algorithm to obtain the smoothed weights, which are expressed as follows: In the formula, Indicates how fast the weight decreases. The smaller the value, the faster the weight decreases. is the EMA value at time k+1, that is, the final weight at time k+1; is the recursive weight at time k+1; Represents the EMA value at time k; Normalize the smoothing weight to get the final weight, which is expressed as follows: 。 6. The method for predicting electricity prices based on dynamic integrated self-adaptation according to claim 1, characterized in that: In the step of inputting the test data into multiple base learner models and combining the final weights to obtain the final prediction value, the final prediction value expression is as follows: In the formula, represents the final weight; Represents the prediction results within multiple base learner models.

7. A power price prediction system based on dynamic integrated self-adaptation, characterized in that: include: A data preprocessing module is used to obtain multivariate electricity price data and preprocess the multivariate electricity price data to obtain a training data set; The model processing module is used to build a dynamic integrated adaptive electricity price prediction model, train multiple base learner models using the training data set, evaluate the performance of multiple base learner models using the validation set, dynamically adjust the weight of the model combination, and obtain the final weight; The model test output module is used to input the test data into multiple base learner models and combine the final weights to obtain the final prediction value.

8. The power price prediction system based on dynamic integrated self-adaptation according to claim 7, characterized in that: The model processing module includes a training module and a verification module; A training module, used to take multivariate electricity price data as input variables and train multiple base learner models through key features of the input variables; The verification module is used to calculate the prediction error using the verification data obtained by evaluating the performance of multiple base learner models using the verification set, and dynamically adjust the weight of the model combination according to the prediction error to obtain the final weight.

9. A mobile terminal comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method for predicting electricity price based on dynamic integrated adaptation as described in any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for predicting electricity price based on dynamic integrated adaptation as described in any one of claims 1 to 6 are implemented.

Images