A multi-day electricity price prediction method, system, electronic device and storage medium

Abstract

This disclosure provides a multi-day electricity price forecasting method, system, electronic device, and storage medium. It employs a hybrid model combining the time-series processing advantages of LSTM networks and the nonlinear forecasting capabilities of XGBoost models to address the highly nonlinear and time-series characteristics of electricity market prices. By dividing the overall forecasting task into multiple time periods and using horizontally cascaded forecasting models to handle different forecasting time ranges, each model focuses on price forecasting for a specific forecasting period. Selecting the optimal model to output the final forecast result allows for more sensitive identification of abnormal fluctuations in price data and a rapid response when price fluctuations occur. The application of unique extreme value theory and change point analysis improves the forecasting accuracy and stability, especially for peak electricity prices, under price fluctuation conditions. Furthermore, the framework is simple, easy to implement and maintain, and reduces forecasting complexity and cost.

Description

Technical Field

[0001] The embodiments disclosed herein belong to the field of electricity price forecasting technology, specifically relating to a multi-day electricity price forecasting method, system, electronic device, and storage medium. Background Technology

[0002] In electricity market price forecasting, with the development of smart grids and big data technologies, an increasing number of electricity price forecasting methods have been proposed to improve forecast accuracy and real-time performance. These methods mainly include statistical analysis methods based on historical data, machine learning algorithms, and deep learning models. However, although these methods have made some progress in many aspects of electricity price forecasting technology, they still have many shortcomings, especially in the area of ​​peak price detection.

[0003] Statistical analysis methods, such as ARIMA (Autoregressive Integrated Moving Average model), are traditional time series forecasting tools. They predict future trends by analyzing past price data. However, because they assume the data is stationary and cannot adapt to sudden price fluctuations, ARIMA's accuracy is significantly reduced when faced with price spikes.

[0004] Machine learning algorithms, such as Support Vector Machines (SVM) and Random Forests (RF), while possessing good predictive capabilities, often fail to perform quickly enough when dealing with extreme fluctuations in electricity prices. Because these algorithms typically rely on large amounts of historical samples to identify patterns, they may exhibit predictive delays when faced with novel price spikes.

[0005] Deep learning models, such as Long Short-Term Memory (LSTM) networks, can capture complex patterns in long-term sequences; however, their complexity and computational cost make them difficult to generalize for real-time applications. Furthermore, the "black box" nature of deep learning models makes their results difficult to interpret, a significant drawback in electricity markets where transparent decision-making is required.

[0006] In summary, traditional statistical methods cannot adapt to the rapidly changing market conditions in a short period of time, resulting in slow response and insufficient timeliness; existing machine learning algorithms have large errors in peak detection, insufficient accuracy, and lack sensitivity to handle emergencies, leading to economic losses; the complex architecture of some deep learning models makes implementation and maintenance costs high, reducing their feasibility for application in small and medium-sized power companies, and the computational requirements and data processing capabilities of specific algorithms lead to high costs, limiting their widespread application. Summary of the Invention

[0007] The embodiments disclosed herein are intended to at least address one of the technical problems existing in the prior art, and to provide a multi-day electricity price forecasting method, system, electronic device, and storage medium.

[0008] One aspect of this disclosure provides a multi-day electricity price forecasting method, the method comprising:

[0009] Obtain historical electricity price data and map the historical electricity price data to a feature space;

[0010] The prediction task is divided into multiple time periods, and different feature subsets of the feature space are assigned to each time period.

[0011] Based on each of the aforementioned feature subsets, multiple horizontally cascaded long short-term memory / extreme gradient boosting hybrid models are used to output the preliminary long short-term memory electricity price predictions and the preliminary extreme gradient boosting electricity price predictions for each of the aforementioned time periods.

[0012] Based on the comparison between the preliminary electricity price forecast value from the Long Short-Term Memory (LSTM) and the preset peak electricity price threshold, the final electricity price forecast value is selected.

[0013] Integrate the final electricity price forecasts for all time periods and output the electricity price forecast results for the total forecast period.

[0014] Further, the step of selecting the final electricity price forecast based on the comparison result between the initial electricity price forecast value from the long short-term memory and the preset peak electricity price threshold includes:

[0015] When the initial electricity price forecast value of the long short-term memory is less than the preset first peak electricity price threshold, the extreme gradient boosting of the initial electricity price forecast value is selected as the final electricity price forecast value.

[0016] When the preliminary electricity price forecast value of the long short-term memory is greater than or equal to the preset first peak electricity price threshold and less than the preset second peak electricity price threshold, the preliminary electricity price forecast value of the long short-term memory is selected as the final electricity price forecast value.

[0017] When the initial electricity price forecast value of the long short-term memory exceeds the preset second peak electricity price threshold, the extreme gradient boosting of the initial electricity price forecast value is selected as the final electricity price forecast value.

[0018] Wherein, the first peak electricity price threshold is less than the second peak electricity price threshold.

[0019] Furthermore, when the preliminary electricity price forecast value of the long short-term memory exceeds the preset second peak electricity price threshold, an early warning signal is issued.

[0020] Furthermore, the prediction task is divided into multiple time periods, including:

[0021] The prediction task is divided into five time periods; wherein, the five time periods include the first hour, the second and third hours, the fourth hour, the fifth to the 24th hour, and the 25th to the 96th hour.

[0022] Another aspect of this disclosure provides a multi-day electricity price forecasting system, characterized in that the system comprises:

[0023] The data acquisition module is used to acquire historical electricity price data and map the historical electricity price data to a feature space;

[0024] The task partitioning module is used to divide the prediction task into multiple time periods and assign different feature subsets of the feature space to each time period.

[0025] The preliminary prediction module is used to output the preliminary electricity price prediction values ​​for each time period and the preliminary electricity price prediction values ​​for each time period through multiple horizontally cascaded long short-term memory / extreme gradient boosting hybrid models based on each of the feature subsets.

[0026] The final prediction module is used to select the final electricity price prediction value based on the comparison result between the initial electricity price prediction value of the long short-term memory and the preset peak electricity price threshold.

[0027] The results output module integrates the final electricity price forecast values ​​for all time periods and outputs the electricity price forecast results for the total forecast period.

[0028] Furthermore, the final prediction module is specifically used for:

[0029] When the initial electricity price forecast value of the long short-term memory is less than the preset first peak electricity price threshold, the extreme gradient boosting of the initial electricity price forecast value is selected as the final electricity price forecast value.

[0030] When the preliminary electricity price forecast value of the long short-term memory is greater than or equal to the preset first peak electricity price threshold and less than the preset second peak electricity price threshold, the preliminary electricity price forecast value of the long short-term memory is selected as the final electricity price forecast value.

[0031] When the initial electricity price forecast value of the long short-term memory exceeds the preset second peak electricity price threshold, the extreme gradient boosting of the initial electricity price forecast value is selected as the final electricity price forecast value.

[0032] Wherein, the first peak electricity price threshold is less than the second peak electricity price threshold.

[0033] Furthermore, the system also includes an early warning signal module, which issues an early warning signal when the preliminary electricity price forecast value of the long short-term memory exceeds a preset second peak electricity price threshold.

[0034] Furthermore, the task partitioning module is specifically used for:

[0035] The prediction task is divided into five time periods; wherein, the five time periods include the first hour, the second and third hours, the fourth hour, the fifth to the 24th hour, and the 25th to the 96th hour.

[0036] Another aspect of this disclosure provides an electronic device, comprising:

[0037] At least one processor; and,

[0038] A memory communicatively connected to the at least one processor is used to store one or more programs that, when executed by the at least one processor, enable the at least one processor to implement the multi-day electricity price forecasting method described above.

[0039] Another aspect of this disclosure provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the multi-day electricity price forecasting method described above.

[0040] This disclosure discloses a multi-day electricity price forecasting method, system, electronic device, and storage medium. It employs a hybrid model combining the time-series processing advantages of LSTM networks and the nonlinear forecasting capabilities of XGBoost models to address the highly nonlinear and time-series characteristics of electricity market prices. By dividing the overall forecasting task into multiple time periods and using horizontally cascaded forecasting models to handle different forecasting time ranges, each model focuses on price forecasting for a specific forecasting time period. Selecting the optimal model to output the final forecast result allows for more sensitive identification of abnormal fluctuations in price data and a rapid response when price fluctuations occur. The application of unique extreme value theory and change point analysis improves the forecasting accuracy and stability, especially for peak electricity prices, under price fluctuation conditions. Furthermore, the framework is simple, easy to implement and maintain, and reduces forecasting complexity and cost. Attached Figure Description

[0041] Figure 1This is a flowchart illustrating a multi-day electricity price forecasting method according to an embodiment of the present disclosure;

[0042] Figure 2 This is a schematic diagram of the horizontal cascade structure of multiple long short-term memory / extreme gradient boosting hybrid models according to another embodiment of this disclosure;

[0043] Figure 3 This is a schematic diagram of the structure of a multi-day electricity price forecasting system according to another embodiment of this disclosure;

[0044] Figure 4 This is a schematic diagram of the structure of an electronic device according to another embodiment of the present disclosure. Detailed Implementation

[0045] The technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. Based on the embodiments of this disclosure, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this disclosure.

[0046] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced without one or more of the specific details, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0047] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0048] It should be understood that although the terms first, second, third, etc., may be used in this disclosure to describe various components, these components should not be limited by these terms. These terms are used to distinguish one component from another. Therefore, the first component discussed below may be referred to as the second component without departing from the teachings of this disclosure. As used in this disclosure, the term "and / or" includes all combinations of any and more of the associated listed items.

[0049] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of exemplary embodiments, and the modules or processes in the drawings are not necessarily necessary for implementing this disclosure, and therefore cannot be used to limit the scope of protection of this disclosure.

[0050] like Figure 1 As shown, one embodiment of this disclosure provides a multi-day electricity price forecasting method, the method comprising:

[0051] Step S1: Obtain historical electricity price data and map the historical electricity price data to the feature space.

[0052] Specifically, time-series electricity price data is obtained from a historical electricity price database, using p t Let p represent the electricity price per hour (t). t Mapped to the feature space X = {x} defined in this embodiment j Let x = {j = 1, 2, ..., N}, where x = ... j This represents the j-th feature variable. The feature space X contains the input variables (features) used for model training and prediction.

[0053] Step S2: Divide the prediction task into multiple time periods and assign different feature subsets of the feature space to each time period.

[0054] Specifically, the prediction task in this embodiment consists of five prediction time periods k. i The system is composed of units i = 1, 2, 3, 4, 5, representing different time periods. Specifically, k1 represents the first hour of the future, k2 represents the second and third hours, k3 represents the fourth hour, k4 represents the fifth to the 24th hour, and k5 represents the 25th to the 96th hour. That is, k... i The corresponding prediction time steps are k1=1, k2∈{2,3}, k3=4, k4∈{5,…,24}, and k5∈{25,…,96}. The time steps for each time period k... i Use a suitable subset of features from the feature space X.

[0055] Step S3: Based on each of the feature subsets, output the preliminary long short-term memory electricity price prediction and the preliminary extreme gradient boosting electricity price prediction for each time period through multiple horizontally cascaded long short-term memory / extreme gradient boosting hybrid models.

[0056] Specifically, the hybrid model in this embodiment consists of a vertically stacked LSTM (Long Short-Term Memory) network and an XGBoost (Extreme Gradient Boosting) tree. For example... Figure 2 As shown, multiple long short-term memory / extreme gradient boosting hybrid models E i The numbers i = 1, 2, 3, 4, 5 are cascaded horizontally, each responsible for a different time period k. i The forecasting task, through the collaborative work of various models, achieves multi-day forecasting of electricity prices over the entire forecast period. For each pre-defined forecast period k... i Each hybrid model uses its corresponding feature subset. As input, output the preliminary long short-term memory electricity price forecast for that time period independently. and extreme gradients to improve preliminary electricity price forecasts Finally, corresponding to each time point t+k i A total of 96 outputs and 96

[0057] LSTM networks are an improved type of recurrent neural network (RNN) suitable for processing and predicting time series data. They can effectively capture the long-term dependencies and sequence information of the data and are often used to solve long-term and short-term memory problems in time series forecasting.

[0058] XGBoost is a high-efficiency Gradient Boosting Decision Tree (GBDT) algorithm that generates a strong predictive model by combining multiple weak learners. It is particularly suitable for capturing the nonlinear features of data and has advantages such as handling missing values, fast training, and high prediction accuracy.

[0059] Step S4: Select the final electricity price prediction value based on the comparison result between the preliminary electricity price prediction value of the long short-term memory and the preset peak electricity price threshold.

[0060] Specifically, a first peak electricity price threshold T1 representing the rightmost (maximum) value of the normal price range is set, and a second peak electricity price threshold T2 representing the super price peak is set. In some embodiments, T2 = 2T1.

[0061] A price spike refers to a sudden and significant increase in electricity market prices, while a super price spike refers to an extreme situation where electricity market prices far exceed the normal range.

[0062] When the initial electricity price forecast for long and short term memory When the price is less than the preset first peak electricity price threshold T1, an extreme gradient is selected to increase the initial electricity price forecast. As of the current time point t+k i Final electricity price forecast When the initial electricity price forecast for long and short term memory When the price is greater than or equal to the preset first peak electricity price threshold T1 and less than the preset second peak electricity price threshold T2, the preliminary electricity price forecast value from the long short-term memory is selected. As of the current time point t+k i Final electricity price forecast When the initial electricity price forecast for long and short term memory When the price exceeds the preset second peak electricity price threshold T2, an extreme gradient is selected to increase the initial electricity price forecast. As of the current time point t+k i Final electricity price forecast It also issues a warning signal to remind relevant technical personnel to take special measures that may be necessary in response to this situation.

[0063] Due to their model characteristics, LSTM excels at capturing long-term trends in time series, but is not sensitive enough to predicting extreme values ​​(super price spikes); XGBoost excels at capturing nonlinear relationships and local features, and its prediction results are relatively more stable when there are significant differences in the data distribution. When XGBoost predicts more accurately on stationary data, its output is chosen; when... At this time, LSTM is more sensitive to capturing short-term price fluctuations, therefore the output of LSTM is selected; when In some cases, the occurrence of super price spikes may correspond to rare events, and XGBoost is more robust to extreme values, so the output of XGBoost is chosen.

[0064] Step S5: Integrate the final electricity price forecast values ​​for all time periods and output the electricity price forecast results for the total forecast period.

[0065] Specifically, the various final electricity price forecasts obtained in the previous step S4 will be used to... By summarizing and integrating the data, we obtain a matrix of electricity price forecasts for the total forecast period, i.e., the next 96 hours.

[0066]

[0067] In some embodiments, combining steps S4 and S5 above, the final electricity price forecast result can be expressed as follows:

[0068]

[0069] In the formula, This is the electricity price forecast matrix. This is the preliminary electricity price forecast for XGBoost. T1 represents the initial electricity price forecast value from the LSTM, T2 represents the first peak electricity price threshold, and T2 represents the second peak electricity price threshold.

[0070] This disclosure discloses a multi-day electricity price forecasting method that employs a hybrid model combining the time-series processing advantages of LSTM networks and the nonlinear forecasting capabilities of XGBoost models to address the highly nonlinear and time-series characteristics of electricity market prices. By dividing the overall forecasting task into multiple time periods and using multiple horizontally cascaded forecasting models to handle different forecasting time periods, each model focuses on electricity price forecasting for a specific forecasting time period. The optimal model is selected to output the final forecast result, enabling more sensitive identification of abnormal fluctuations in price data and rapid response when price fluctuations occur. The application of unique extreme value theory and change point analysis improves the forecasting accuracy and stability, especially for peak electricity prices, under price fluctuation conditions. Furthermore, the framework is simple, easy to implement and maintain, and reduces forecasting complexity and cost.

[0071] like Figure 3 As shown, another embodiment of this disclosure provides a multi-day electricity price forecasting system, the system comprising:

[0072] Data acquisition module 310 is used to acquire historical electricity price data and map the historical electricity price data to a feature space;

[0073] The task partitioning module 320 is used to divide the prediction task into multiple time periods and assign different feature subsets of the feature space to each time period.

[0074] The preliminary prediction module 330 is used to output the preliminary electricity price prediction values ​​for each time period and the preliminary electricity price prediction values ​​for each time period through multiple horizontally cascaded long short-term memory / extreme gradient boosting hybrid models based on each of the feature subsets.

[0075] The final prediction module 340 is used to select the final electricity price prediction value based on the comparison result between the initial electricity price prediction value of the long short-term memory and the preset peak electricity price threshold.

[0076] The output module 350 is used to integrate the final electricity price forecast values ​​for all time periods and output the electricity price forecast results for the total forecast period.

[0077] For example, the final prediction module 340 is specifically used for:

[0078] When the initial electricity price forecast value of the long short-term memory is less than the preset first peak electricity price threshold, the extreme gradient boosting of the initial electricity price forecast value is selected as the final electricity price forecast value.

[0079] When the preliminary electricity price forecast value of the long short-term memory is greater than or equal to the preset first peak electricity price threshold and less than the preset second peak electricity price threshold, the preliminary electricity price forecast value of the long short-term memory is selected as the final electricity price forecast value.

[0080] When the initial electricity price forecast value of the long short-term memory exceeds the preset second peak electricity price threshold, the extreme gradient boosting of the initial electricity price forecast value is selected as the final electricity price forecast value.

[0081] Wherein, the first peak electricity price threshold is less than the second peak electricity price threshold.

[0082] For example, such as Figure 3 As shown, the system also includes an early warning signal module 360, which is used to issue an early warning signal when the preliminary electricity price forecast value of the long short-term memory is greater than the preset second peak electricity price threshold.

[0083] For example, the task partitioning module 320 is specifically used for:

[0084] The prediction task is divided into five time periods; wherein, the five time periods include the first hour, the second and third hours, the fourth hour, the fifth to the 24th hour, and the 25th to the 96th hour.

[0085] Specifically, a multi-day electricity price forecasting system according to an embodiment of this disclosure is used to implement the multi-day electricity price forecasting method described in the above embodiments. The specific implementation process has been described in detail in the above embodiments and will not be repeated here.

[0086] This disclosure discloses a multi-day electricity price forecasting system that employs a hybrid model combining the time-series processing advantages of LSTM networks and the nonlinear forecasting capabilities of XGBoost models to address the highly nonlinear and time-series characteristics of electricity market prices. By dividing the overall forecasting task into multiple time periods and using multiple horizontally cascaded forecasting models to handle different forecasting time periods, each model focuses on electricity price forecasting for a specific forecasting time period. The system selects the optimal model to output the final forecast result, enabling more sensitive identification of abnormal fluctuations in price data and rapid response when price fluctuations occur. The application of unique extreme value theory and change point analysis improves the forecasting accuracy and stability, especially for peak electricity prices, under price fluctuation conditions. Furthermore, the system has a simple framework, is easy to implement and maintain, and reduces forecasting complexity and cost.

[0087] like Figure 4 As shown, another embodiment of this disclosure provides an electronic device, including:

[0088] At least one processor 401; and a memory 402 communicatively connected to the at least one processor 401 for storing one or more programs that, when executed by the at least one processor 401, enable the at least one processor 401 to implement the multi-day electricity price forecasting method described above.

[0089] The memory 402 and processor 401 are connected via a bus, which may include any number of interconnecting buses and bridges. The bus connects various circuits of one or more processors 401 and memory 402 together. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. A bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by processor 401 is transmitted over a wireless medium via an antenna, which further receives data and transmits it to processor 401.

[0090] Processor 401 is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory 402 can be used to store data used by processor 401 during operation.

[0091] Another embodiment of this disclosure provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the multi-day electricity price forecasting method described above.

[0092] The computer-readable storage medium may be included in the systems or electronic devices disclosed herein, or it may exist independently.

[0093] Computer-readable storage media can be any tangible medium that contains or stores a program, and can be an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device. More specific examples include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, optical fibers, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0094] Computer-readable storage media may also include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code, specific examples of which include, but are not limited to, electromagnetic signals, optical signals, or any suitable combination thereof.

[0095] It is understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of this disclosure, and this disclosure is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and substance of this disclosure, and these modifications and improvements are also considered to be within the scope of protection of this disclosure.

Claims

1. A multi-day electricity price forecasting method, characterized in that, The method includes: Obtain historical electricity price data and map the historical electricity price data to a feature space; The prediction task is divided into multiple time periods, and different feature subsets of the feature space are assigned to each time period. Based on each of the aforementioned feature subsets, multiple horizontally cascaded long short-term memory / extreme gradient boosting hybrid models are used to output the preliminary long short-term memory electricity price predictions and the preliminary extreme gradient boosting electricity price predictions for each of the aforementioned time periods. Based on the comparison between the preliminary electricity price forecast value from the Long Short-Term Memory (LSTM) and the preset peak electricity price threshold, the final electricity price forecast value is selected. Integrate the final electricity price forecasts for all time periods and output the electricity price forecast results for the total forecast period.

2. The method according to claim 1, characterized in that, The step of selecting the final electricity price forecast based on the comparison between the initial electricity price forecast value from the Long Short-Term Memory (LSTM) and the preset peak electricity price threshold includes: When the initial electricity price forecast value of the long short-term memory is less than the preset first peak electricity price threshold, the extreme gradient boosting of the initial electricity price forecast value is selected as the final electricity price forecast value. When the preliminary electricity price forecast value of the long short-term memory is greater than or equal to the preset first peak electricity price threshold and less than the preset second peak electricity price threshold, the preliminary electricity price forecast value of the long short-term memory is selected as the final electricity price forecast value. When the initial electricity price forecast value of the long short-term memory exceeds the preset second peak electricity price threshold, the extreme gradient boosting of the initial electricity price forecast value is selected as the final electricity price forecast value. Wherein, the first peak electricity price threshold is less than the second peak electricity price threshold.

3. The method according to claim 2, characterized in that, When the preliminary electricity price forecast value of the long short-term memory exceeds the preset second peak electricity price threshold, an early warning signal is issued.

4. The method according to any one of claims 1 to 3, characterized in that, The process of dividing the prediction task into multiple time periods includes: The prediction task is divided into five time periods; wherein, the five time periods include the first hour, the second and third hours, the fourth hour, the fifth to the 24th hour, and the 25th to the 96th hour.

5. A multi-day electricity price forecasting system, characterized in that, The system includes: The data acquisition module is used to acquire historical electricity price data and map the historical electricity price data to a feature space; The task partitioning module is used to divide the prediction task into multiple time periods and assign different feature subsets of the feature space to each time period. The preliminary prediction module is used to output the preliminary electricity price prediction values ​​for each time period and the preliminary electricity price prediction values ​​for each time period through multiple horizontally cascaded long short-term memory / extreme gradient boosting hybrid models based on each of the feature subsets. The final prediction module is used to select the final electricity price prediction value based on the comparison result between the initial electricity price prediction value of the long short-term memory and the preset peak electricity price threshold. The results output module integrates the final electricity price forecast values ​​for all time periods and outputs the electricity price forecast results for the total forecast period.

6. The system according to claim 5, characterized in that, The final prediction module is specifically used for: When the initial electricity price forecast value of the long short-term memory is less than the preset first peak electricity price threshold, the extreme gradient boosting of the initial electricity price forecast value is selected as the final electricity price forecast value. When the preliminary electricity price forecast value of the long short-term memory is greater than or equal to the preset first peak electricity price threshold and less than the preset second peak electricity price threshold, the preliminary electricity price forecast value of the long short-term memory is selected as the final electricity price forecast value. When the initial electricity price forecast value of the long short-term memory exceeds the preset second peak electricity price threshold, the extreme gradient boosting of the initial electricity price forecast value is selected as the final electricity price forecast value. Wherein, the first peak electricity price threshold is less than the second peak electricity price threshold.

7. The system according to claim 6, characterized in that, The system also includes an early warning signal module, which issues an early warning signal when the preliminary electricity price forecast value of the long short-term memory exceeds the preset second peak electricity price threshold.

8. The system according to any one of claims 5 to 7, characterized in that, The task partitioning module is specifically used for: The prediction task is divided into five time periods; wherein, the five time periods include the first hour, the second and third hours, the fourth hour, the fifth to the 24th hour, and the 25th to the 96th hour.

9. An electronic device, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor is used to store one or more programs that, when executed by the at least one processor, enable the at least one processor to implement the multi-day electricity price forecasting method according to any one of claims 1 to 4.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the multi-day electricity price forecasting method according to any one of claims 1 to 4.

Images