Large user power load analysis method and system based on time sequence model

A three-layer architecture with a time series model improves power load forecasting accuracy and real-time performance by effectively processing and synchronizing large-user data, addressing inefficiencies in existing methods.

CN119089338APending Publication Date: 2024-12-06STATE GRID SHANGHAI MUNICIPAL ELECTRIC POWER CO +1
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202410975364.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

Existing power load forecasting methods face challenges with insufficient data, low model accuracy, and poor real-time performance, particularly in large-user scenarios, due to ineffective utilization of historical and real-time data without effective data mining.

Method used

A three-layer architecture comprising a load data feature engineering layer, an analysis and prediction model layer, and a K-line application service layer, utilizing a time series model (ARIMA) to process and synchronize data, eliminate errors, and generate intelligent load analysis charts across various devices.

Benefits of technology

Enhances data utilization and accuracy, providing real-time and dynamic power load forecasting with efficient data query and prediction services, supporting diverse business scenarios and dimensions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119089338A_ABST
    Figure CN119089338A_ABST
Patent Text Reader

Abstract

The invention relates to a large user power load analysis method based on a time sequence model, and the method comprises the following steps: a load data feature engineering step: collecting power load curve data and user file data, and carrying out the data processing and cleaning, and obtaining load data features; an analysis and prediction step: performing data synchronization and storage on the obtained load data characteristics, combining real-time data with historical data, and performing data analysis on the combined load data characteristics to obtain a data analysis result; and a K-line application service step: generating a load K-line analysis chart according to the obtained data analysis result, and displaying the load K-line analysis chart through terminal equipment. Compared with the prior art, the method has the advantages that effective utilization of massive power load big data is realized, the processing flow of the big data is optimized, comprehensive prediction is performed by integrating real-time data and historical data, and the accuracy and the real-time performance of load prediction are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of power load forecasting, and in particular to a large-user power load analysis method and system based on a time series model. Background Art

[0002] Load analysis in power systems is an important issue in the power industry. Traditional power load analysis methods usually rely on historical data and empirical models, which may face problems such as insufficient data, low model accuracy, and poor real-time performance. With the development of big data technology, power load analysis methods based on big data can more effectively utilize massive data and improve the accuracy and real-time performance of load forecasting.

[0003] The invention with publication number CN116702955A discloses a circuit load prediction method based on an electric power big data mining algorithm, comprising the following steps: S1: preprocessing the load data; S2: discretizing the load data based on the X-Means algorithm to determine the load range; S3: using the random forest algorithm to establish a classification model to predict the power load range for the next day or month; S4: using the random forest algorithm to establish a regression model to predict the specific load value for the next day or month; S5: combining the time series model in WEKA with the random forest method to perform time series prediction on historical load data to predict the specific load value at the future moment.

[0004] Most existing solutions directly perform load forecasting on preprocessed load data without effective data mining, resulting in heavy data processing load and ineffective use of real-time and historical data. Summary of the invention

[0005] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a large-user power load analysis method and system based on a time series model, which can more effectively utilize massive data and improve the accuracy and real-time performance of load forecasting.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] A method for analyzing large-user power load based on a time series model comprises the following steps:

[0008] Load data feature engineering steps: Collect power load curve data and user profile data, and process and clean the data to obtain load data features;

[0009] Analysis and prediction step: synchronize and store the acquired load data characteristics, combine the real-time data with the historical data, analyze the combined load data characteristics, and obtain the data analysis results;

[0010] K-line application service steps: Generate a load K-line analysis chart based on the acquired data analysis results and display it through the terminal device.

[0011] Furthermore, the collected power load curve data is a 96-point load curve of the user, and the data processing and cleaning process includes: eliminating the user's failed collection data; using the user's previous day's load average value to fill in the missing values ​​and abnormal values ​​collected on the day.

[0012] Furthermore, the load data characteristics include normalized maximum, minimum and average values ​​of user daily load and corresponding user profile data.

[0013] Furthermore, the user profile data includes basic user information, industry type, industrial type and power supply company information.

[0014] Furthermore, the data analysis process includes industry electricity load analysis, industrial electricity consumption characteristics analysis and regional electricity consumption characteristics analysis.

[0015] Furthermore, the data analysis process uses an ARIMA algorithm model to generate a time series from load data characteristics, thereby performing analysis and prediction.

[0016] Furthermore, a T+2 scheme is used to extract data to realize data synchronization function. The T+2 scheme is that the synchronization of actual business data in the production environment and business data in the K-line application is 2 days apart.

[0017] The present invention also provides a power load analysis system based on big data, comprising:

[0018] The load data feature engineering layer is used to collect power load curve data and user profile data, and perform data processing and cleaning to obtain load data features;

[0019] The analysis and prediction model layer is used to synchronize and store the acquired load data characteristics, combine the real-time data with the historical data, perform data analysis on the combined load data characteristics, and obtain the data analysis results;

[0020] The K-line application layer is used to generate a load K-line analysis chart based on the acquired data analysis results and display it through the terminal device.

[0021] Furthermore, the load data feature engineering layer uses the power marketing collection system to obtain the user's 96-point load curve as power load curve data, and obtains user profile data; removes user collection failure data; uses the user's previous day load average value to fill in the missing values ​​and collection abnormal values ​​of the day; generates the normalized user daily load maximum value, minimum value and average value and corresponding user profile data for the processed data;

[0022] The user profile data includes basic user information, industry type, industrial type and power supply company information;

[0023] The analysis and prediction model layer adopts the T+2 scheme to extract data and realize the data synchronization function. The T+2 scheme is that the actual business data of the production environment and the business data in the K-line application are synchronized with each other for 2 days; the load data characteristics acquired in real time are combined with historical data for data storage; the load data characteristics that are combined with user profile data and power load curve data are converted into a time series using the ARIMA algorithm model, so as to perform analysis and prediction, and complete industry electricity load analysis, industrial electricity characteristics analysis and regional electricity characteristics analysis.

[0024] Furthermore, the K-line application layer displays the generated load K-line analysis chart through a PC, a mobile device or a large screen.

[0025] Compared with the prior art, the present invention has the following advantages:

[0026] (1) The system of the present invention includes a load data feature engineering layer, an analysis and prediction model layer, and a K-line application service layer, with a unified technical architecture; it involves business entities and business processes related to power loads, and the basic business data and link data have been fully covered, which can meet the needs of future business scenario construction and smart application construction.

[0027] (2) The load data feature engineering layer in the present invention uses the user's 96-point load data to construct load feature engineering, forming features such as maximum value, minimum value, instantaneous value and average value. In this process, the actual collection business is combined to eliminate the failed collection data, and the previous day's load average value is used to fit and fill in the collection missing values, thereby realizing the user load curve feature engineering, which can meet the data requirements of the analysis and prediction model.

[0028] (3) The analysis and prediction model layer in the present invention utilizes power load data and combines it with the ARIMA algorithm model to complete the rapid conversion from massive data to simple business models, and from complex and cumbersome business logic to a simple and abstract power load business data model, thereby realizing dynamic and efficient business data query and providing an agile and fast business data service interface and prediction service interface for K-line application services.

[0029] (4) The K-line application service layer in the present invention provides a customizable software client, which can realize human-computer interaction through various terminal devices such as PCs, mobile devices, and large screens. It has a high degree of intelligence and is convenient for data analysis and forecast reporting for load services in various dimensions. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1A schematic diagram of data flow processing of a large-user power load analysis method based on a time series model provided in an embodiment of the present invention;

[0031] Figure 2 A schematic flow chart of a large-user power load analysis method based on a time series model provided in an embodiment of the present invention;

[0032] Figure 3 This is an architecture diagram of a power load analysis system based on big data provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions 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 part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0034] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0035] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0036] Example 1

[0037] like Figure 1 and Figure 2 As shown, this embodiment provides a large user power load analysis method based on a time series model, comprising the following steps:

[0038] Load data feature engineering steps: Collect power load curve data and user profile data, and process and clean the data to obtain load data features;

[0039] Analysis and prediction step: synchronize and store the acquired load data characteristics, combine the real-time data with the historical data, analyze the combined load data characteristics, and obtain the data analysis results;

[0040] K-line application service steps: Generate a load K-line analysis chart based on the acquired data analysis results and display it through the terminal device.

[0041] The specific steps are as follows:

[0042] S1: Obtain user profile data and power load curve data;

[0043] The power marketing collection system can be used to obtain the user's 96-point load curve and comprehensive multiples. The user profile data includes basic user information, industry type, industrial type and power supply company information;

[0044] S2: Delete the load data that failed to be collected according to actual business requirements;

[0045] S3: Use the average load value of the previous day to fill in the missing load values ​​of the current day and collect abnormal values;

[0046] S4: construct the maximum value, minimum value, average value, instantaneous value and other characteristic values, and normalize them;

[0047] The T+2 solution is used to extract data and realize data synchronization. The T+2 solution means that the actual business data in the production environment and the business data in the K-line application are synchronized 2 days apart. The load data characteristics acquired in real time are combined with historical data for data storage;

[0048] S5: Input feature data into ARIMA model training and verification to generate analysis and prediction model;

[0049] The ARIMA algorithm model is used to generate a time series from the load data characteristics that combine user profile data and power load curve data, so as to conduct analysis and prediction, and complete industry power load analysis, industrial power consumption characteristics analysis and regional power consumption characteristics analysis.

[0050] S6: Generate load K-line analysis display.

[0051] It provides a customizable software client that can realize human-computer interaction through various terminal devices such as PC, mobile devices, large screens, etc. It has a high degree of intelligence and is convenient for data analysis and forecast reporting for load businesses in various dimensions.

[0052] like Figure 3 As shown, this embodiment also provides a power load analysis system based on big data, including:

[0053] The load data feature engineering layer is used to collect power load curve data and user profile data, and perform data processing and cleaning to obtain load data features;

[0054] The analysis and prediction model layer is used to synchronize and store the acquired load data characteristics, combine the real-time data with the historical data, perform data analysis on the combined load data characteristics, and obtain the data analysis results;

[0055] The K-line application layer is used to generate a load K-line analysis chart based on the acquired data analysis results and display it through the terminal device.

[0056] Specifically, the load data feature engineering layer uses the power marketing collection system to obtain the user's 96-point load curve as the power load curve data, and obtains the user's profile data; eliminates the user's failed collection data; uses the user's average load value of the previous day to fill in the missing values ​​and abnormal collection values ​​of the current day; generates the normalized user's daily load maximum, minimum and average values ​​and the corresponding user profile data for the processed data; realizes the engineering of user load curve features, which can meet the data requirements of the analysis and prediction model.

[0057] User profile data includes basic user information, industry type, industrial type and power supply company information;

[0058] The analysis and prediction model layer uses the T+2 scheme to extract data and realize the data synchronization function. The T+2 scheme is that the actual business data in the production environment and the business data in the K-line application are synchronized 2 days apart; the load data characteristics acquired are combined with historical data for data storage; the load data characteristics combined with user profile data and power load curve data are used to generate time series using the ARIMA algorithm model, so as to perform analysis and prediction, and complete industry power load analysis, industrial power characteristics analysis and regional power characteristics analysis;

[0059] In this embodiment, the analysis and prediction model layer includes a power load analysis and prediction model design scheme and a power load data storage synchronization scheme, the power load analysis and prediction model design scheme includes a simple abstract unit for the energy meter acquisition business process and a data analysis unit, the power load data storage synchronization scheme includes a data storage unit and a data synchronization unit that are interconnected, and the data storage unit and the data synchronization unit are interconnected;

[0060] The data analysis unit conducts data analysis specifically including:

[0061] Combine power consumption archive information and collected load data to conduct industry power load analysis, industrial power consumption characteristics analysis and regional power consumption characteristics analysis;

[0062] The data synchronization unit uses the T+2 solution to extract data and implement data synchronization. The T+2 solution means that the actual business data in the production environment and the business data in the K-line application are synchronized 2 days apart.

[0063] Business processing includes collecting load business, user peak, flat and valley electricity consumption and metering device event information business;

[0064] The data processed by the business includes the user's historical electricity usage information;

[0065] The forecasting model uses the ARIMA model to generate time series and provide a forecasting function interface.

[0066] The K-line application layer displays the generated load K-line analysis chart through PC, mobile device or large screen.

[0067] Specifically, the K-line application service layer provides a customizable software client that can achieve human-computer interaction through various terminal devices such as PCs, mobile devices, and large screens. It has a high degree of intelligence and is convenient for data analysis and forecast reporting for load businesses in various dimensions.

[0068] In summary, the present invention adopts a three-layer architecture technical solution, the bottom layer is the load data feature engineering layer, the load data feature engineering layer, uses the user's 96-point load data to construct load feature engineering, and forms features such as maximum value, minimum value, instantaneous value and average value. In this process, the actual collection business is combined to eliminate the failed collection data, and the previous day's load average value is used to fit and fill the collection missing value, thereby realizing the user load curve feature engineering, which can meet the data requirements of the analysis and prediction model; the middle is the analysis and prediction model layer, which uses the power load data and combines the ARIMA algorithm model to complete the rapid conversion from massive data to simple business models, and transforms from complex and cumbersome business logic to a simple and abstract power load business data model, realizing dynamic and efficient business data query, and providing agile and fast business data service interface and prediction service interface for K-line application services; the top layer is the K-line application service layer, which provides a customizable software client, which can realize human-computer interaction through various terminal devices such as PC, mobile devices, and large screens, with a high degree of intelligence, and is convenient for data analysis and prediction reporting for load businesses of various dimensions.

[0069] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A method for analyzing power load of large users based on a time series model, characterized in that: The following steps are involved: Load data feature engineering steps: Collect power load curve data and user profile data, and process and clean the data to obtain load data features; Analysis and prediction step: synchronize and store the acquired load data characteristics, combine the real-time data with the historical data, analyze the combined load data characteristics, and obtain the data analysis results; K-line application service steps: Generate a load K-line analysis chart based on the acquired data analysis results and display it through the terminal device.

2. According to the method for analyzing large user power load based on time series model in claim 1, it is characterized in that: The collected power load curve data is a 96-point load curve of the user. The data processing and cleaning process includes: removing the user's failed collection data; using the user's load average value of the previous day to fill in the missing values ​​and abnormal values ​​collected on the current day.

3. The large-user power load analysis method based on a time series model according to claim 1 is characterized in that: The load data characteristics include the normalized maximum, minimum and average values ​​of the user's daily load and the corresponding user profile data.

4. The method for analyzing power load of large users based on a time series model according to claim 1 is characterized in that: The user profile data includes basic user information, industry type, industrial type and power supply company information.

5. The method for analyzing power load of large users based on a time series model according to claim 1 is characterized in that: The data analysis process includes industry electricity load analysis, industrial electricity consumption characteristics analysis and regional electricity consumption characteristics analysis.

6. The method for analyzing power load of large users based on a time series model according to claim 1 is characterized in that: The data analysis process uses the ARIMA algorithm model to generate a time series from the load data characteristics, thereby performing analysis and prediction.

7. The method for analyzing power load of large users based on a time series model according to claim 1 is characterized in that: The T+2 scheme is used to extract data and realize the data synchronization function. The T+2 scheme is that the actual business data in the production environment and the business data in the K-line application are synchronized 2 days apart.

8. A power load analysis system based on big data, characterized in that: include: The load data feature engineering layer is used to collect power load curve data and user profile data, and perform data processing and cleaning to obtain load data features; The analysis and prediction model layer is used to synchronize and store the acquired load data characteristics, combine the real-time data with the historical data, analyze the combined load data characteristics, and obtain the data analysis results; The K-line application layer is used to generate a load K-line analysis chart based on the acquired data analysis results and display it through the terminal device.

9. The power load analysis system based on big data according to claim 8, characterized in that: The load data feature engineering layer uses the power marketing collection system to obtain the user's 96-point load curve as power load curve data, and obtains user profile data; Eliminate user data that failed to be collected; use the average value of the user's load on the previous day to fill in the missing values ​​and abnormal values ​​collected on the current day; generate the normalized maximum, minimum and average values ​​of the user's daily load and the corresponding user profile data for the processed data; The user profile data includes basic user information, industry type, industrial type and power supply company information; The analysis and prediction model layer adopts the T+2 scheme to extract data and realize the data synchronization function. The T+2 scheme is that the actual business data of the production environment and the business data in the K-line application are synchronized with each other for 2 days; the load data characteristics acquired in real time are combined with historical data for data storage; the load data characteristics that are combined with user profile data and power load curve data are converted into a time series using the ARIMA algorithm model, so as to perform analysis and prediction, and complete industry electricity load analysis, industrial electricity characteristics analysis and regional electricity characteristics analysis.

10. The power load analysis system based on big data according to claim 8, characterized in that: The K-line application layer displays the generated load K-line analysis chart through a PC, a mobile device or a large screen.

Citation Information

Patent Citations

  • Circuit load prediction method based on electric power big data mining algorithm

    CN116702955A

  • Electrical load management method and system for industrial enterprise aggregate user

    CN106651200A