Power business data processing method and system
By using power supply prediction models and power consumption prediction models in microgrid systems, the problem of difficult power supply and demand balance in the existing technology is solved, and more accurate power supply and demand change prediction and power grid scheduling is achieved, and the stability of the system is improved. reliability.
Patent Information
- Application Number
- CN202411265898.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-11
- Publication Date
- 2025-02-14
AI Technical Summary
The existing microgrid systems lack efficient power service data processing capabilities, especially in real-time data acquisition, accurate prediction and intelligent scheduling, making it difficult to achieve a balance between power supply and demand.
Through the power supply prediction model and power consumption prediction model, the power supply data and local power consumption data for the future period are predicted, combined with the local power consumption peak and power storage data, the overall power supply and maximum power consumption are predicted, thereby achieving more accurate power Forecast of supply and demand changes and grid scheduling.
It realizes more accurate prediction of power supply and demand changes, improves the accuracy of grid scheduling, and enhances the stability and reliability of the power system.
Smart Images
Figure CN119443339A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of data processing, and in particular to a method and system for processing power business data. Background Art
[0002] In the current power system, microgrid technology has attracted widespread attention as an important means to improve energy efficiency and grid reliability. A microgrid is a small, controllable local power grid system that usually includes power generation, energy storage and loads. It can operate independently of the traditional power grid or in parallel with it. Although microgrid technology has many advantages, such as improving energy utilization efficiency, enhancing grid stability and supporting the integration of renewable energy, it still faces some challenges in practical applications.
[0003] Microgrid systems in existing technologies often lack efficient power business data processing capabilities, especially in real-time data collection, accurate prediction and intelligent scheduling. These systems find it difficult to achieve the optimal balance between power supply and demand when dealing with the volatility and uncertainty of renewable energy and changing load demands. Summary of the invention
[0004] The present invention predicts the power supply data and local power consumption data for a period of time in the future through a power supply prediction model and a power consumption prediction model, and then predicts the overall power supply and possible maximum power consumption for a period of time in the future based on the local power consumption peak and electric energy storage data, thereby more accurately predicting changes in power supply and demand and achieving more precise power grid scheduling.
[0005] A method for processing power business data, comprising:
[0006] S1: obtaining real-time power consumption data, real-time meteorological data and real-time power supply data in the current monitoring period, and storing the real-time power consumption data, real-time meteorological data and real-time power supply data in the power business database;
[0007] S2: compare the real-time power consumption data with the real-time power supply data. If the real-time power supply data is greater than the real-time power consumption data, calculate the difference between the real-time power consumption data and the real-time power supply data to obtain the excess power supply data; determine whether the excess power supply data is greater than the electric energy storage threshold. If the excess power supply data is greater than the electric energy storage threshold, store the electric energy corresponding to the part of the excess power supply data that exceeds the electric energy storage threshold in the electric energy storage model; otherwise, do nothing;
[0008] Obtain the electric energy storage data in the electric energy storage model at the current time; if the real-time power supply data is less than the real-time power consumption data, calculate the difference between the real-time power consumption data and the real-time power supply data to obtain the power supply shortage data; determine whether the electric energy storage data is greater than the power supply shortage data, if the electric energy storage data is greater than the power supply shortage data, and the difference between the electric energy storage data and the power supply shortage data is greater than the electric energy buffer threshold, use the electric energy storage model to supply power to the user; otherwise, enable the power grid to supply power to the user;
[0009] S3: Obtain the historical power supply data and the corresponding historical meteorological data within the T time before the current monitoring time point, and simultaneously obtain the meteorological forecast data within the current prediction target time period K, combine the meteorological forecast data, the historical power supply data and the corresponding historical meteorological data into power supply characteristic data, input the power supply characteristic data into the power supply prediction model, and obtain the power supply prediction data within the current prediction target time period K;
[0010] S4: For the area with regular power consumption, obtain the historical local power consumption data in the T time before the current monitoring time point, input the historical local power consumption data into the power consumption prediction model, and obtain the local power consumption prediction data in the current prediction target time period K;
[0011] S5: Obtain the local peak power consumption in the area with frequently changing power consumption in the current regular cycle, add the local peak power consumption to the local power consumption forecast data, and obtain the highest predicted power consumption data within the current forecast target time period K; add the currently obtained electric energy storage data to the power supply forecast data, and obtain the overall predicted power supply data within the current forecast target time period K; determine whether the highest predicted power consumption data is greater than the overall predicted power supply data. If the highest predicted power consumption data is greater than the overall predicted power supply data, enable the power grid to supply power to users; otherwise, no operation is performed.
[0012] Preferably, the specific operation of marking the area with constant power consumption change and the area with regular power consumption includes:
[0013] A1: Set a regular cycle and set the area division threshold to divide a day into several time periods;
[0014] A2: Divide the electricity consumption area into several local areas, obtain the electricity consumption data of each local area in each time period of each day in the previous regular cycle of the current regular cycle, obtain the electricity consumption data sets of several time periods, calculate the standard deviation of the electricity consumption data set of each time period, and calculate the average of the standard deviations of all time periods to obtain the average standard deviation value of each local area;
[0015] A3: For any local area, if the average standard deviation of the local area is greater than the area division threshold, the local area is marked as a power consumption constant change area; otherwise, the area is marked as a power consumption regular area;
[0016] A4: Execute steps A2-A3 once in each regular cycle to mark the power consumption constant change area and the power consumption regular area of the current regular cycle.
[0017] Preferably, obtaining the local peak power consumption of the area with constant power consumption in the current regular cycle includes:
[0018] The previous regular cycle of the current regular cycle is divided into several time periods with a time interval of K, and the time period power consumption data of the area with constant power consumption in all time periods are obtained respectively. The time period power consumption data with the highest value is taken as the local power consumption peak value of the area with constant power consumption in the current regular cycle.
[0019] Preferably, the power supply prediction model is established based on the LSTM model, including Q power supply prediction LSTM units, 1 first fully connected layer and 1 first output layer, wherein the power supply prediction LSTM unit is used to extract the timing features in the power supply characteristic data; the first fully connected layer is used to extract features from the outputs of all power supply prediction LSTM units; and the first output layer is used to output the power supply prediction data.
[0020] Preferably, the training of the power supply prediction model specifically includes the following operations:
[0021] Acquire several historical meteorological data sets within a time period of T+K, wherein the historical meteorological data sets contain Q+1 meteorological data arranged in chronological order, wherein T=KQ, and mark the power supply data of the corresponding time for all meteorological data, and acquire several power supply prediction training samples, and use the first Q meteorological data in the power supply prediction training samples as the input data of the power supply prediction model, and use the last meteorological data in the power supply prediction training samples as the output data of the power supply prediction model; divide all power supply prediction training samples into a power supply prediction training set and a power supply prediction test set, use the power supply prediction training set to train the power supply prediction model, and then input the power supply prediction test set into the power supply prediction model to obtain a first test result; set a first training condition, and if the first test result meets the first training condition, output the trained power supply prediction model; otherwise, continue to use the power supply prediction training set to train the power supply prediction model.
[0022] Preferably, the electricity consumption prediction model is established based on the LSTM model, including Q electricity consumption prediction LSTM units, 1 second fully connected layer and 1 second output layer, wherein the electricity consumption prediction LSTM unit is used to extract the temporal features in the historical local electricity consumption data; the second fully connected layer is used to extract features from the outputs of all electricity consumption prediction LSTM units; and the second output layer is used to output local electricity consumption prediction data.
[0023] Preferably, the training of the electricity consumption prediction model specifically includes the following operations:
[0024] Acquire several sets of historical electricity consumption data in a time period of T+K for a region with regular electricity consumption, the historical electricity consumption data set contains Q+1 sets of electricity consumption data arranged in chronological order, use the first Q sets of electricity consumption data in the historical electricity consumption data set as the input of the electricity consumption prediction model, and use the last set of electricity consumption data in the historical electricity consumption data set as the output of the electricity consumption prediction model; divide all historical electricity consumption data sets into an electricity consumption prediction training set and an electricity consumption prediction test set, use the electricity consumption prediction training set to train the electricity consumption prediction model, and then input the electricity consumption prediction test set into the electricity consumption prediction model to obtain a second test result; set a second training condition, and if the second test result meets the second training condition, output the trained electricity consumption prediction model; otherwise, continue to use the electricity consumption prediction training set to train the electricity consumption prediction model.
[0025] Preferably, a power business data processing system comprises:
[0026] The power business data acquisition module is used to obtain real-time power consumption data, real-time meteorological data and real-time power supply data during the current monitoring period;
[0027] The power business database module is used to store real-time power consumption data, real-time meteorological data and real-time power supply data in the power business database;
[0028] The electric energy storage module is used to store the electric energy corresponding to the part of the excess power supply data exceeding the electric energy storage threshold in the electric energy storage model when the excess power supply data is greater than the electric energy storage threshold; and to supply power to the user using the electric energy storage model when the electric energy storage data is greater than the power shortage data and the difference between the electric energy storage data and the power shortage data is greater than the electric energy buffer threshold;
[0029] The power supply prediction module is used to input the power supply characteristic data into the power supply prediction model to obtain the power supply prediction data within the current prediction target time period K;
[0030] The power consumption prediction module is used to input the historical local power consumption data into the power consumption prediction model to obtain the local power consumption prediction data within the current prediction target time period K;
[0031] The prediction response module is used to determine whether the highest predicted power consumption data is greater than the overall predicted power supply data. If the highest predicted power consumption data is greater than the overall predicted power supply data, the power grid is enabled to supply power to the user.
[0032] The present invention has the following advantages:
[0033] 1. The present invention compares the real-time electricity consumption data and the real-time power supply data. If the real-time power supply data is greater than the real-time electricity consumption data, and the value by which the real-time power supply data is greater than the real-time electricity consumption data is higher than the electric energy storage threshold, the electric energy corresponding to the part of the excess power supply data that exceeds the electric energy storage threshold is stored in the electric energy storage model, thereby realizing dynamic adjustment of electric energy storage.
[0034] 2. The present invention predicts the power supply data and local power consumption data for a period of time in the future through the power supply prediction model and the power consumption prediction model, and then predicts the overall power supply and possible maximum power consumption for a period of time in the future based on the local power consumption peak and electric energy storage data, thereby more accurately predicting changes in power supply and demand and achieving more accurate power grid scheduling. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] Figure 1 This is a schematic diagram of the structure of the power business data processing system used in an embodiment of the present invention. DETAILED DESCRIPTION
[0036] In order to enable persons skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.
[0037] Embodiment 1, a method for processing power business data, comprising:
[0038] S1: obtain the real-time electricity consumption data, real-time meteorological data and real-time power supply data of the current monitoring period, and store the real-time electricity consumption data, real-time meteorological data and real-time power supply data in the power business database. The real-time electricity consumption data includes the power consumption of users during the monitoring period. The real-time meteorological data has a significant impact on renewable energy power generation. The real-time power supply data is used to express the amount of electricity generated by renewable energy during the current monitoring period. The power business database is a specially designed storage system for storing various data related to the power business for subsequent processing and analysis.
[0039] S2: Compare the real-time power consumption data with the real-time power supply data. If the real-time power supply data is greater than the real-time power consumption data, calculate the difference between the real-time power consumption data and the real-time power supply data to obtain the excess power supply data. This comparison process is crucial to maintaining the balance between power supply and demand, because excessive power supply may lead to energy waste or equipment overload; determine whether the excess power supply data is greater than the power storage threshold. If the excess power supply data is greater than the power storage threshold, store the power corresponding to the part of the excess power supply data that exceeds the power storage threshold in the power storage model. The power storage threshold is used to determine when the excess power should be stored; otherwise, no operation; storage operation helps to optimize the operation of the microgrid. By storing excess power, it can be released during peak power consumption hours, thereby balancing the load of the microgrid;
[0040] Obtain the electric energy storage data in the electric energy storage model at the current time; if the real-time power supply data is less than the real-time power consumption data, indicating that there is a power shortage, then calculate the difference between the real-time power consumption data and the real-time power supply data to obtain the power shortage data; determine whether the electric energy storage data is greater than the power shortage data, if the electric energy storage data is greater than the power shortage data, and the difference between the electric energy storage data and the power shortage data is greater than the electric energy buffer threshold, the electric energy buffer threshold is used to determine whether the stored electric energy is needed to supplement the power shortage, then use the electric energy storage model to supply power to the user; otherwise, enable the power grid to supply power to the user to ensure that the user's power demand is met;
[0041] S3: Obtain the historical power supply data and corresponding historical meteorological data within the T time before the current monitoring time point, T can be 24 hours, and obtain the meteorological forecast data within the current prediction target time period K, K can be one hour, combine the meteorological forecast data, historical power supply data and corresponding historical meteorological data into power supply characteristic data, input the power supply characteristic data into the power supply prediction model, and obtain the power supply prediction data within the current prediction target time period K. The power supply prediction data provides the predicted value of future power supply, which is crucial for microgrid scheduling and energy management;
[0042] S4: For the area with regular power consumption, obtain the historical local power consumption data in the T time before the current monitoring time point, input the historical local power consumption data into the power consumption prediction model, and obtain the local power consumption prediction data in the current prediction target time period K. The local power consumption prediction data can be used for microgrid planning and operation, and is also helpful for load management, and optimizes the use of power resources through demand response and other measures;
[0043] S5: Obtain the local peak power consumption of the area with constant power consumption in the current regular cycle. The regular cycle can be one month, which is determined according to the actual situation. Add the local peak power consumption to the local power consumption forecast data to obtain the highest predicted power consumption data in the current forecast target time period K. The highest predicted power consumption data represents the possible power consumption of the entire power consumption area within the future K time period; add the currently obtained electric energy storage data to the power supply forecast data to obtain the overall predicted power supply data within the current forecast target time period K. The overall predicted power supply data represents the total amount of electric energy generated by renewable energy and stored by the electric energy storage model within the future K time period; determine whether the highest predicted power consumption data is greater than the overall predicted power supply data. If the highest predicted power consumption data is greater than the overall predicted power supply data, enable the power grid to supply power to users. This operation is to prevent the area with constant power consumption from suddenly generating higher power consumption, resulting in insufficient power supply for the microgrid; otherwise, no operation is performed. If the highest predicted power consumption data is not greater than the overall predicted power supply data, it means that the current power supply capacity can meet the predicted power demand, and no additional power supply measures are required.
[0044] The specific operations for marking the areas with frequent power consumption changes and areas with regular power consumption include:
[0045] A1: Set a regular cycle and set a regional division threshold. The regional division threshold is used to distinguish between areas with constant power consumption and areas with regular power consumption. This threshold is usually based on statistical metrics such as standard deviation or rate of change to evaluate the stability of power consumption. Divide a day into several time periods.
[0046] A2: Divide the electricity consumption area into several local areas, obtain the electricity consumption data of each local area in each time period of each day in the previous regular cycle of the current regular cycle, obtain the electricity consumption data sets of several time periods, and calculate the standard deviation of the electricity consumption data set of each time period. The standard deviation is a statistical indicator to measure the size of data fluctuations and reflects the stability of electricity consumption. Then calculate the average value of the standard deviation of all time periods to obtain the average standard deviation value of each local area, which will be used to evaluate the stability of electricity consumption in the entire local area;
[0047] A3: For any local area, if the average standard deviation of the local area is greater than the regional division threshold, it indicates that the power consumption in the area fluctuates greatly, and the local area is marked as a power consumption constant fluctuation area; otherwise, it indicates that the power consumption in the area is relatively stable, and the area is marked as a power consumption regular area;
[0048] A4: Execute steps A2-A3 once in each regular cycle to mark the areas with frequently changing power consumption and areas with regular power consumption in the current regular cycle. By periodically executing these steps, the system can continuously monitor and update the marks of areas with frequently changing power consumption and areas with regular power consumption. As time goes by and power consumption patterns change, periodic execution can ensure that the system adapts to new power consumption trends and adjusts area marks in a timely manner.
[0049] Obtain the local peak power consumption in the current regular cycle in the area where power consumption changes frequently, including:
[0050] The previous regular cycle of the current regular cycle is divided into several time periods with a time interval of K, and the time period power consumption data of the area with constant power consumption in all time periods are obtained respectively. The time period power consumption data with the highest value is taken as the local power consumption peak value of the area with constant power consumption in the current regular cycle. The local power consumption peak value can be used for system load forecasting, microgrid planning and operation scheduling, and help predict and prepare for power consumption peaks.
[0051] The power supply prediction model is established based on the LSTM model. LSTM is a special type of recurrent neural network, which is very suitable for processing and predicting time series data. LSTM can learn long-term dependencies in data, which makes it very effective in scenarios such as power load forecasting. The power supply prediction model includes Q power supply prediction LSTM units, 1 first fully connected layer and 1 first output layer, where the power supply prediction LSTM unit is used to extract the temporal features in the power supply feature data; the first fully connected layer is used to extract features from the outputs of all power supply prediction LSTM units; the first output layer is used to output power supply prediction data.
[0052] The training of the power supply prediction model includes the following operations:
[0053] A number of historical meteorological data sets within a time period of T+K are obtained, wherein the historical meteorological data sets contain Q+1 meteorological data arranged in chronological order, wherein T=KQ, i.e., a time period of T contains Q time periods of K. The power supply data of the corresponding time are marked for all meteorological data, so that the model can learn the relationship between meteorological data and power supply. A number of power supply prediction training samples are obtained, and the first Q meteorological data in the power supply prediction training samples are used as input data of the power supply prediction model, and the last meteorological data in the power supply prediction training samples are used as output data of the power supply prediction model. All power supply prediction training samples are divided into a power supply prediction training set and a power supply prediction test set, and the power supply prediction model is trained using the power supply prediction training set, and then the power supply prediction test set is input into the power supply prediction model to obtain a first test result. A first training condition is set, which may be some standards about model performance, such as mean square error or accuracy. If the first test result meets the first training condition, it means that the model has been trained well enough and can be used for actual power supply prediction, and the trained power supply prediction model is output; otherwise, the power supply prediction training set is continued to be used to train the power supply prediction model.
[0054] The electricity consumption prediction model is established based on the LSTM model, including Q electricity consumption prediction LSTM units, 1 second fully connected layer and 1 second output layer. The electricity consumption prediction LSTM unit is used to extract the temporal features in the historical local electricity consumption data; the second fully connected layer is used to extract features from the outputs of all electricity consumption prediction LSTM units; and the second output layer is used to output local electricity consumption prediction data.
[0055] The training of the electricity consumption prediction model includes the following operations:
[0056] Acquire several sets of historical electricity consumption data in a time period of T+K for a region with regular electricity consumption, the historical electricity consumption data set contains Q+1 sets of electricity consumption data arranged in chronological order, use the first Q sets of electricity consumption data in the historical electricity consumption data set as the input of the electricity consumption prediction model, and use the last set of electricity consumption data in the historical electricity consumption data set as the output of the electricity consumption prediction model; divide all historical electricity consumption data sets into an electricity consumption prediction training set and an electricity consumption prediction test set, use the electricity consumption prediction training set to train the electricity consumption prediction model, and then input the electricity consumption prediction test set into the electricity consumption prediction model to obtain a second test result; set a second training condition, and if the second test result meets the second training condition, output the trained electricity consumption prediction model; otherwise, continue to use the electricity consumption prediction training set to train the electricity consumption prediction model.
[0057] Embodiment 2, a power business data processing system, such as Figure 1 As shown, including:
[0058] The power business data acquisition module is used to obtain real-time power consumption data, real-time meteorological data and real-time power supply data during the current monitoring period;
[0059] The power business database module is used to store real-time power consumption data, real-time meteorological data and real-time power supply data in the power business database;
[0060] The electric energy storage module is used to store the electric energy corresponding to the part of the excess power supply data exceeding the electric energy storage threshold in the electric energy storage model when the excess power supply data is greater than the electric energy storage threshold; and to supply power to the user using the electric energy storage model when the electric energy storage data is greater than the power shortage data and the difference between the electric energy storage data and the power shortage data is greater than the electric energy buffer threshold;
[0061] The power supply prediction module is used to input the power supply characteristic data into the power supply prediction model to obtain the power supply prediction data within the current prediction target time period K;
[0062] The power consumption prediction module is used to input the historical local power consumption data into the power consumption prediction model to obtain the local power consumption prediction data within the current prediction target time period K;
[0063] The prediction response module is used to determine whether the highest predicted power consumption data is greater than the overall predicted power supply data. If the highest predicted power consumption data is greater than the overall predicted power supply data, the power grid is enabled to supply power to the user.
[0064] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the appended claims of the present invention. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.
Claims
1. A method for processing power business data, characterized in that: include: S1: obtaining real-time power consumption data, real-time meteorological data and real-time power supply data in the current monitoring period, and storing the real-time power consumption data, real-time meteorological data and real-time power supply data in the power business database; S2: comparing the real-time power consumption data with the real-time power supply data. If the real-time power supply data is greater than the real-time power consumption data, calculating the difference between the real-time power consumption data and the real-time power supply data to obtain the excess power supply data. Determine whether the excess power supply data is greater than the electric energy storage threshold, and if the excess power supply data is greater than the electric energy storage threshold, store the electric energy corresponding to the portion of the excess power supply data exceeding the electric energy storage threshold in the electric energy storage model; Otherwise, no operation; Obtain the electric energy storage data in the electric energy storage model at the current time; If the real-time power supply data is less than the real-time power consumption data, the difference between the real-time power consumption data and the real-time power supply data is calculated to obtain the power supply shortage data; determine whether the electric energy storage data is greater than the power supply shortage data; if the electric energy storage data is greater than the power supply shortage data, and the difference between the electric energy storage data and the power supply shortage data is greater than the electric energy buffer threshold, the electric energy storage model is used to supply power to the user; otherwise, the power grid is enabled to supply power to the user; S3: Obtain the historical power supply data and the corresponding historical meteorological data within the T time before the current monitoring time point, and simultaneously obtain the meteorological forecast data within the current prediction target time period K, combine the meteorological forecast data, the historical power supply data and the corresponding historical meteorological data into power supply characteristic data, input the power supply characteristic data into the power supply prediction model, and obtain the power supply prediction data within the current prediction target time period K; S4: For the area with regular power consumption, obtain the historical local power consumption data in the T time before the current monitoring time point, input the historical local power consumption data into the power consumption prediction model, and obtain the local power consumption prediction data in the current prediction target time period K; S5: Obtain the local peak power consumption of the area with constant power consumption in the current regular cycle, add the local peak power consumption to the local power consumption forecast data, and obtain the highest predicted power consumption data in the current forecast target time period K; add the currently obtained electric energy storage data to the power supply forecast data, and obtain the overall predicted power supply data in the current forecast target time period K; Determine whether the highest predicted power consumption data is greater than the overall predicted power supply data. If the highest predicted power consumption data is greater than the overall predicted power supply data, enable the power grid to supply power to the user; otherwise, no operation is performed.
2. A method for processing power business data according to claim 1, characterized in that: The specific operations for marking the areas with frequent power consumption changes and areas with regular power consumption include: A1: Set a regular cycle and set the area division threshold to divide a day into several time periods; A2: Divide the electricity consumption area into several local areas, obtain the electricity consumption data of each local area in each time period of each day in the previous regular cycle of the current regular cycle, obtain the electricity consumption data sets of several time periods, calculate the standard deviation of the electricity consumption data set of each time period, and calculate the average of the standard deviations of all time periods to obtain the average standard deviation value of each local area; A3: For any local area, if the average standard deviation of the local area is greater than the area division threshold, the local area is marked as a power consumption constant change area; otherwise, the area is marked as a power consumption regular area; A4: Execute steps A2-A3 once in each regular cycle to mark the power consumption constant change area and the power consumption regular area of the current regular cycle.
3. A method for processing power business data according to claim 2, characterized in that: Obtain the local peak power consumption in the current regular cycle in the area where power consumption changes frequently, including: The previous regular cycle of the current regular cycle is divided into several time periods with a time interval of K, and the time period power consumption data of the area with constant power consumption in all time periods are obtained respectively. The time period power consumption data with the highest value is taken as the local power consumption peak value of the area with constant power consumption in the current regular cycle.
4. A method for processing power business data according to claim 3, characterized in that: The power supply prediction model is established based on the LSTM model, including Q power supply prediction LSTM units, 1 first fully connected layer and 1 first output layer. The power supply prediction LSTM unit is used to extract the timing features in the power supply feature data; the first fully connected layer is used to extract features from the outputs of all power supply prediction LSTM units; and the first output layer is used to output the power supply prediction data.
5. A method for processing power business data according to claim 4, characterized in that: The training of the power supply prediction model includes the following operations: Obtain several historical meteorological data sets within a time period of T+K, where the historical meteorological data sets contain Q+1 meteorological data arranged in chronological order, where T=KQ, and mark the power supply data of the corresponding time for all meteorological data, obtain several power supply prediction training samples, use the first Q meteorological data in the power supply prediction training samples as input data of the power supply prediction model, and use the last meteorological data in the power supply prediction training samples as output data of the power supply prediction model; divide all power supply prediction training samples into a power supply prediction training set and a power supply prediction test set, use the power supply prediction training set to train the power supply prediction model, and then input the power supply prediction test set into the power supply prediction model to obtain a first test result; A first training condition is set. If the first test result meets the first training condition, the trained power supply prediction model is output; otherwise, the power supply prediction model is continuously trained using the power supply prediction training set.
6. A method for processing power business data according to claim 5, characterized in that: The electricity consumption prediction model is established based on the LSTM model, including Q electricity consumption prediction LSTM units, 1 second fully connected layer and 1 second output layer. The electricity consumption prediction LSTM unit is used to extract the temporal features in the historical local electricity consumption data; the second fully connected layer is used to extract features from the outputs of all electricity consumption prediction LSTM units; and the second output layer is used to output local electricity consumption prediction data.
7. A method for processing power business data according to claim 6, characterized in that: The training of the electricity consumption prediction model includes the following operations: Obtain several sets of historical electricity consumption data in a time period of T+K for a region with regular electricity consumption, wherein the historical electricity consumption data set contains Q+1 sets of electricity consumption data arranged in chronological order, and the first Q sets of electricity consumption data in the historical electricity consumption data set are used as inputs of the electricity consumption prediction model, and the last set of electricity consumption data in the historical electricity consumption data set is used as outputs of the electricity consumption prediction model; divide all historical electricity consumption data sets into an electricity consumption prediction training set and an electricity consumption prediction test set, train the electricity consumption prediction model using the electricity consumption prediction training set, and then input the electricity consumption prediction test set into the electricity consumption prediction model to obtain a second test result; A second training condition is set. If the second test result satisfies the second training condition, the trained power consumption prediction model is output; otherwise, the power consumption prediction training set is continued to be used to train the power consumption prediction model.
8. A power business data processing system, characterized in that: The system is used for a power business data processing method as described in any one of claims 1 to 7, comprising: The power business data acquisition module is used to obtain real-time power consumption data, real-time meteorological data and real-time power supply data during the current monitoring period; The power business database module is used to store real-time power consumption data, real-time meteorological data and real-time power supply data in the power business database; The electric energy storage module is used to store the electric energy corresponding to the part of the excess power supply data exceeding the electric energy storage threshold in the electric energy storage model when the excess power supply data is greater than the electric energy storage threshold; and to supply power to the user using the electric energy storage model when the electric energy storage data is greater than the power shortage data and the difference between the electric energy storage data and the power shortage data is greater than the electric energy buffer threshold; The power supply prediction module is used to input the power supply characteristic data into the power supply prediction model to obtain the power supply prediction data within the current prediction target time period K; The power consumption prediction module is used to input the historical local power consumption data into the power consumption prediction model to obtain the local power consumption prediction data within the current prediction target time period K; The prediction response module is used to determine whether the highest predicted power consumption data is greater than the overall predicted power supply data. If the highest predicted power consumption data is greater than the overall predicted power supply data, the power grid is enabled to supply power to the user.