User electric energy load prediction and regulation method and system based on agent interaction
By combining user intelligent agent interaction technology with electrical energy and production scheduling data, the problems of single data and insufficient user participation in existing power load forecasting methods have been solved, realizing efficient and accurate load forecasting and control, and improving user work efficiency and information interaction convenience.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-25
- Publication Date
- 2026-03-31
AI Technical Summary
Existing power load forecasting methods suffer from limitations such as a single data dimension, insufficient integration of key data like user production schedules, and a lack of active user participation in the forecasting process. This results in low forecasting accuracy, high implementation costs, and susceptibility to the inaccuracy of metering data.
By adopting a user-based intelligent agent interaction method, combining user-side power energy collection data, production scheduling data, and marketing operation platform data, the load forecasting process is planned through intelligent agent technology and natural language interaction with administrators. MCP technology is used for forecasting, and user active participation is introduced to improve forecasting accuracy and convenience.
It enables more accurate electricity load forecasting, reduces implementation costs, improves user work efficiency, reduces communication costs, provides a more flexible and natural way of information interaction, and improves the accuracy of forecasts and users' ability to perceive changes in production.
Smart Images
Figure CN121769845A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power technology, and specifically to a method and system for predicting and regulating user power load based on intelligent agent interaction. Background Technology
[0002] With the deepening of electricity market reforms and the development of new electricity business models such as virtual power plants, higher requirements are being placed on the accuracy of user electricity load forecasting. Accurate electricity load forecasting not only provides data support for users to participate in electricity market transactions and optimize production scheduling, but also ensures the safe and stable operation of virtual power plants and the power grid.
[0003] In existing technologies, electricity load forecasting methods mainly include forecasting methods based on intelligent agent simulation, forecasting methods based on big data mining, and forecasting methods based on statistical analysis. For example, some patents use intelligent agent technology to simulate individual user behavior and combine it with the Monte Carlo method for load forecasting; other patents use K-Means clustering algorithms, deep learning, and other technologies to make forecasts based on historical electricity consumption data, marketing data, etc.
[0004] However, these existing technologies all have certain shortcomings. For example, they rely on a single data dimension, mostly considering only user-side electricity energy collection data, marketing data, or meteorological data, without fully incorporating key data that directly affects electricity load changes, such as user production scheduling. Some technologies lack active user participation in the forecasting process, making it difficult to accurately capture load fluctuations caused by unexpected events such as adjustments to user production plans and equipment maintenance. In addition, some methods rely on complex mathematical models or deep learning algorithms, resulting in high implementation costs and demanding hardware requirements. Furthermore, some methods based on statistical values or historical data are susceptible to the accuracy of metering data, leading to significant forecasting bias.
[0005] Therefore, there is a need to provide a power load forecasting method and system that can enrich data dimensions, introduce active user participation, and balance forecasting accuracy with ease of implementation. Summary of the Invention
[0006] The technical problem to be solved by this invention is to address the aforementioned deficiencies in the existing technology by providing a method and system for predicting and regulating user power load based on user-intelligent agent interaction. This method utilizes partial or full data from user-side power energy collection data, user-side production scheduling data, marketing and operation platform user data, and user-intelligent agent interaction data to more accurately predict and regulate the power load of a single target user or user group. This provides more accurate data reference for users participating in power market transactions, production scheduling, virtual power plant dispatching, and other scenarios.
[0007] According to a first aspect of the present invention, a method for predicting and regulating user power load based on intelligent agent interaction is provided, comprising: First step: Collect historical electricity load data of target users or user groups from the marketing operations platform through scheduled tasks; The second step is to calculate the average daily electricity consumption data of each user in the collected historical electricity load data to obtain the average daily electricity consumption data of the user. The third step: Sort users using the average daily electricity consumption data, and save the users and their average daily electricity consumption data to the database in the form of key-value pairs; Step 4: Construct a user management agent to interact with the administrator via natural language. Use the user management agent to send daily user management confirmation messages to the administrator and receive confirmation instructions from the administrator regarding user information of users who need special load forecasting and the data types used for load forecasting. Step 5: The user management agent uses the confirmed user information and the data type used for load forecasting as prompt words to plan the load forecasting process and obtain the first forecast user list and the second forecast user list. Step 6: The user management agent uses MCP technology to call the corresponding prediction methods for the first and second predicted user lists respectively.
[0008] Preferably, the users requiring special load forecasting are a predetermined number of users obtained by sorting users using average daily electricity consumption data.
[0009] Preferably, user information includes whether the user's production schedule changes daily or weekly.
[0010] Preferably, the data type used for load forecasting includes key-value pairs stored in a database.
[0011] Preferably, the electricity load forecast for the first predicted user list is made by using user data published on the marketing operations platform, and then the obtained user electricity load forecast is corrected by using user-side power energy collection data; the electricity load forecast for the second predicted user list is made by using user-side power energy collection data, user data collected on the marketing operations platform, and user-side production scheduling data.
[0012] Preferably, the first prediction method is invoked for the first list of predicted users: For a workday, the load calculation formula for each time point within the 24 hours of that day is as follows: Where t represents 24 time points, This represents the user electricity load at time t j weeks prior to weekday i in the data released by the marketing center. If the date i of the previous week is not a weekday, it is set to 0 and the denominator is reduced by 1. This represents the user load at time point t, the first working day before working day i, in the user-side data collection. This represents the user load at time point t, the second nearest working day before working day i, in the user-side data collection. For cases where the day is a non-working day, but not during the Spring Festival or National Day holidays, the load calculation formula for that day is: ,in, This represents the user electricity load at time t, which is the j-th week prior to date i and is a non-working day i in the data released by the marketing center. If date i of the j-th week prior is a working day, then set it to 0 and subtract 1 from the denominator. This represents the user load at time point t, which is the first non-working day before non-working day i in the user-side data collection. For cases where the day falls during the Spring Festival or National Day holiday, the load calculation formula for that day is as follows: , This indicates the electricity load at time t on the same date last year, as published by the marketing center. If the same date last year was not a holiday, then the electricity load at time t on the last day of the same holiday last year is used. This represents the electricity load at time t on the same date last year in the user-side data collection. If the same date last year was not a holiday, then the user's electricity load at time t on the last day of the same holiday last year is used.
[0013] Preferably, the second prediction method is invoked for the second prediction user list; in the second prediction method, based on the context information of the interaction between the user management agent and the manager, it is confirmed whether there are any changes in the production schedule of the user; if there are no changes, the first prediction method is executed for the second prediction user list; if there are changes, the user is guided to provide the time of the change and relevant information on the application's electrical load, and this information is combined into system prompt words, which are then used by the user management agent to infer and generate adjusted electrical load prediction data.
[0014] Preferably, the first and second predicted user lists are obtained by using Skill technology to plan the load forecasting process.
[0015] According to a second aspect of the present invention, a system for predicting and regulating user power load based on intelligent agent interaction is provided, comprising: Data acquisition module: Collects historical electricity load data of target users or user groups from the marketing operations platform through scheduled tasks; Calculation module: Calculates the average daily electricity consumption data of each user in the collected historical electricity load data to obtain the average daily electricity consumption data of the user; Storage module: Sort users using average daily electricity consumption data and save users and their average daily electricity consumption data to the database in the form of key-value pairs; User Management Intelligent Agent Module: Used to interact with the administrator via natural language, send daily user management confirmation messages to the administrator, and receive confirmation instructions from the administrator regarding user information of users who need special load forecasting and the data types used for load forecasting; List Acquisition Module: The user management agent uses the confirmed user information and the data type used for load forecasting as prompt words to plan the load forecasting process and obtain the first and second predicted user lists. Prediction Module: The user management agent uses MCP technology to call the corresponding prediction methods for the first and second prediction user lists respectively.
[0016] This invention utilizes advanced artificial intelligence models and intelligent agent technology to provide a more convenient and natural information interaction method for managing users and target users, helping various users improve the efficiency of their load forecasting work. Intelligent agent technology also provides a more flexible way to deliver data, charts, and other information to users, enhancing the flexibility of information acquisition during user decision-making. Furthermore, this invention combines the traditional use of mathematical, machine learning, and deep learning models for objective user load forecasting with the subjective initiative of users' production conditions, enabling users to "actively" participate in the load forecasting process. This allows for more accurate perception of user production changes and more precise user load forecasting.
[0017] By employing this invention, users can utilize portable mobile devices such as smartphones and tablets, thereby increasing work convenience and improving the work efficiency of target users. Furthermore, by incorporating target users into the user load prediction process, objective prediction methods based on mathematical, statistical, and algorithmic models become proactive, enabling more effective and accurate prediction of user load. This also avoids or reduces significant communication costs with target users. Attached Figure Description
[0018] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0019] Figure 1 This is a schematic diagram of the overall process of a user power load prediction and control method based on intelligent agent interaction according to an embodiment of the present invention.
[0020] Figure 2 This is a flowchart of a method for predicting and regulating user power load based on intelligent agent interaction according to an embodiment of the present invention.
[0021] Figure 3 This is a block diagram of a system for predicting and regulating user power load based on intelligent agent interaction according to an embodiment of the present invention. Detailed Implementation
[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0023] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.
[0024] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.
[0025] Figure 1 This is a schematic diagram of the overall process of a user power load prediction and control method based on intelligent agent interaction according to an embodiment of the present invention. Figure 2 This is a flowchart of a method for predicting and regulating user power load based on intelligent agent interaction according to an embodiment of the present invention. (Reference) Figure 1 and Figure 2 As shown, the method for predicting and regulating user power load based on intelligent agent interaction according to an embodiment of the present invention includes: Step S1: Collect historical electricity load data of target users or user groups from the marketing operations platform through scheduled tasks; preferably, the historical electricity load data covers a time range of at least more than 1 day; Step S2: Calculate the average daily electricity consumption data of each user in the collected historical electricity load data to obtain the average daily electricity consumption data of the user. Step S3: Sort users using average daily electricity consumption data and save the users and their average daily electricity consumption data to the database in the form of key-value pairs; Step S4: Construct a user management agent to interact with the administrator via natural language. Use the user management agent to send daily user management confirmation messages to the administrator (e.g., via SMS, email, APP, or mini-program), and receive confirmation instructions from the administrator regarding user information of users who need special load forecasting and the data type used for load forecasting. For example, users requiring special load forecasting are a predetermined number of users obtained by sorting users using average daily electricity consumption data; user information includes whether users frequently have changes in production schedules, such as whether there are changes in production schedules every day or week; the data types used for load forecasting include key-value pairs stored in the database.
[0026] In other words, the user management agent can provide convenient interaction to user managers (administrators) through natural language. Administrators can use the user management agent to query information such as the number of users whose average daily electricity consumption is among the top predetermined number, and the average daily electricity consumption data of each user, which can be used for subsequent load forecasting of users in each type of predicted user list.
[0027] Step S5: The user management agent uses the confirmed user information and the data type used for load forecasting as prompt words to plan the load forecasting process (e.g., using Skill technology) and obtain the first and second predicted user lists. Step S6: The user management agent uses MCP technology to call the corresponding prediction methods for the first and second predicted user lists respectively.
[0028] Specifically, in a preferred embodiment, user load forecasting for the first predicted user list is performed using user data published by the marketing operations platform. Then, the obtained user load forecasting is corrected using user-side energy collection data (i.e., user load scale correction). User load forecasting for the second predicted user list is performed using user-side energy collection data, user collection data from the marketing operations platform, and user-side production scheduling data.
[0029] Specifically, in a preferred embodiment, a first prediction method is invoked for the first predicted user list: For a workday, the load calculation formula for each time point within the 24 hours of that day is as follows: Where t represents 24 time points, This represents the user electricity load at time t j weeks prior to weekday i in the data released by the marketing center. If the date i of the previous week is not a weekday, it is set to 0 and the denominator is reduced by 1. This represents the user load at time point t, the first working day before working day i, in the user-side data collection. This represents the user load at time point t, the second nearest working day before working day i, in the user-side data collection. For cases where the day is a non-working day, but not during the Spring Festival or National Day holidays, the load calculation formula for that day is: ,in, This represents the user electricity load at time t, which is the j-th week prior to date i and is a non-working day i in the data released by the marketing center. If date i of the j-th week prior is a working day, then set it to 0 and subtract 1 from the denominator. This represents the user load at time point t, which is the first non-working day before non-working day i in the user-side data collection. For cases where the day falls during the Spring Festival or National Day holiday, the load calculation formula for that day is as follows: , This indicates the electricity load at time t on the same date last year, as published by the marketing center. If the same date last year was not a holiday, then the electricity load at time t on the last day of the same holiday last year is used. This represents the electricity load at time t on the same date last year in the user-side data collection. If the same date last year was not a holiday, then the user's electricity load at time t on the last day of the same holiday last year is used.
[0030] The first forecasting method uses only user-side electrical energy collection data and user data published on the sales and operation platform for forecasting. It can be used for users with relatively stable production and no changes in production schedules.
[0031] Specifically, in a preferred embodiment, the second prediction method is invoked for the second predicted user list: Based on the context information from the interaction with the manager, the user management agent sends a production scheduling confirmation request to the target predicted user in the second predicted user list through the MCP interface to confirm whether the user has any changes in production scheduling.
[0032] If there is no change (at this point, for example, the target prediction user can reply with a response such as "no change" or "produce as planned," and the management agent receives the user's prompt), then the first prediction method is executed for the second prediction user list.
[0033] If there are changes, the user is guided to provide the time of the change and relevant information about the application's electrical load. This information is combined into system prompts, and the user management agent infers and generates adjusted electrical load forecast data.
[0034] Specifically, for example, if there are changes, the management agent guides the user to provide specific information about the time of change in the production schedule through prompts such as "What changes have occurred in the production schedule?". Examples include "Equipment maintenance from 2 PM to 4 PM, no production" or "Tomorrow production will be extended to 8 PM, with hourly electricity load during the extended period similar to 2 PM." During the communication process, users can also obtain relevant data or charts through natural language questions. Then, the first prediction method is executed to obtain the first 24-hour prediction data. The system then combines user feedback into prompts, such as "A day has 24 hours, working hours are 8 AM to 6 PM, and the hourly electricity consumption during working hours is 1000 MWh. Today, due to special circumstances, production time needs to be extended to 8 PM. The hourly electricity consumption during the extended period is expected to remain the same as during normal time. Output the hourly electricity consumption data for the 24 hours after the production adjustment, presented in tabular form. (Note that the electricity consumption at 9 PM is the same as the electricity consumption from 8 AM to 9 PM)." This allows the management agent to infer from the user prompts and generate a new electricity load prediction based on the production adjustment.
[0035] It is evident that, in addition to publishing user data and collecting user-side data on the sales and operations platform, the second prediction method also requires collaboration with target users through an intelligent agent system to determine whether there are any changes in production scheduling.
[0036] Finally, all the predicted data can be stored in a database.
[0037] Figure 3 This is a block diagram of a user power load prediction and control system based on intelligent agent interaction according to an embodiment of the present invention. Figure 3 As shown, a user power load prediction and control system based on intelligent agent interaction according to an embodiment of the present invention may include: Data acquisition module 10: Collects historical electricity load data of target users or user groups from the marketing operation platform through scheduled tasks; preferably, the historical electricity load data covers a time range of at least more than 1 day; Calculation module 20: Calculates the average daily electricity consumption data of each user in the collected historical electricity load data to obtain the average daily electricity consumption data of the user; Storage module 30: Sorts users using daily average electricity consumption data and saves the users and their daily average electricity consumption data to the database in the form of key-value pairs; User Management Intelligent Agent Module 40: Used to interact with the administrator through natural language, send daily user management confirmation messages to the administrator, and receive confirmation instructions from the administrator regarding user information of users who need special load forecasting and the data type used for load forecasting; List Acquisition Module 50: The user management agent uses the confirmed user information and the data type used for load forecasting as prompt words to plan the load forecasting process (e.g., using Skill technology) and obtain the first predicted user list and the second predicted user list. Prediction Module 60: The user management agent uses MCP technology to call the corresponding prediction methods for the first and second prediction user lists respectively.
[0038] The specific operation of the user power load prediction and regulation system based on intelligent agent interaction according to an embodiment of the present invention can be as described above.
[0039] In summary, this invention proposes a method for more accurate electricity load prediction for a single target user or user group based on user-side power energy collection data, user-side production scheduling data, marketing operation platform user data, and user-smart agent interaction data.
[0040] At the data level, compared to existing technologies that only consider user-side power energy collection data, marketing center user data, and meteorological data, this invention focuses on adding user-side production scheduling data collection and "proactive intervention" of target predicted users. This is because user-side production scheduling data can more directly reflect changes in user production energy consumption, such as production delays, advances, and equipment maintenance. This allows for more accurate user load prediction and adjustment without using complex algorithms. Moreover, the second prediction method uses user scheduling data, which is slightly different from the traditional method.
[0041] Meanwhile, this invention introduces intelligent agent technology, intelligent agent skill technology, and MCP technology to enable more precise control over intelligent agent process orchestration and the use of external tools. At the same time, the large-scale natural language dialogue and understanding capabilities of the intelligent agent technology can provide users with a more natural and convenient system interaction experience, helping users improve overall work efficiency and reduce the investment of a large amount of hardware equipment costs.
[0042] This invention uses a large-model-based agent construction technique, but it is not used to simulate and predict individual user behavior. Instead, it utilizes the agent's natural language capabilities to enable functions such as information confirmation and input with the user. By introducing the user's active participation process, it avoids the errors caused by simulating and predicting individual user behavior.
[0043] This invention includes user-side production scheduling data, user-agent interaction data, and time-of-use electricity pricing policy data of the user's province. It can also use prediction deviation correction methods based on similar days of electricity consumption and scheduling and based on user-agent interaction data, thereby providing more data dimensions and more accurate predictions.
[0044] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts.
[0045] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.
[0046] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.
[0047] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.
Claims
1. A method for user electric energy load prediction and regulation based on agent interaction, characterized in that Comprising: A first step: collecting historical electricity load data of target users or user groups from a marketing operation platform through a timing task; A second step: calculating the average daily electricity data of each user in the collected historical electricity load data to obtain user daily average electricity scale data; A third step: sorting users using the daily average electricity scale data, and saving the users and their daily average electricity scale data in the form of key-value pairs to a database; A fourth step: constructing a user management agent for interacting with administrators through natural language, using the user management agent to initiate daily user management confirmation information to the administrator, and receiving the administrator's confirmation instructions on the user information and data types for load prediction of users that need to be predicted; A fifth step: the user management agent uses the confirmed user information and data types for load prediction as prompt words to plan the load prediction process, and obtains a first prediction user list and a second prediction user list; A sixth step: the user management agent calls the corresponding prediction method for the first prediction user list and the second prediction user list to make predictions through MCP technology.
2. The method of claim 1, wherein, The users that need to be predicted are the top predetermined number of users obtained after sorting the users using the daily average electricity scale data.
3. The method of claim 1, wherein, The user information includes whether the user has production scheduling changes every day or every week.
4. The method of claim 1, wherein, The data types for load prediction include key-value pairs stored in the database.
5. The method of claim 1, wherein, The marketing operation platform is used to publish user data for user electricity load prediction of the first prediction user list, and then the user side electricity energy collection data is used to correct the obtained user electricity load prediction; the user side electricity energy collection data, marketing operation platform user collection data, and user side production scheduling data are used for user electricity load prediction of the second prediction user list.
6. The method of claim 1, wherein, For the first prediction user list, a first prediction method is called: For the case that the day is a working day, the load calculation formula at each time point in 24 hours of the day is: wherein t represents 24 time points, represents the user electricity load at the time point t of the working day i in the jth week before the time point t of the working day i in the marketing center published data, and is set to 0 if the date i in the jth week is not a working day, and the denominator is reduced by 1; represents the user load at the time point t of the first working day before the working day i in the user side collected data; represents the user load at the time point t of the second nearest working day before the working day i in the user side collected data; For the case that the day is a non-working day, but is not during the Spring Festival or National Day, the load calculation formula for the day is: wherein, represents the user power consumption load at time point t of non-working day i weeks before date i in the marketing center published data, and is set to 0 if the date i weeks before is a working day, and the denominator is reduced by 1; represents the user load at time point t of the first non-working day before non-working day i in the user side collected data; For cases where the day falls during the Spring Festival or National Day holiday, the load calculation formula for that day is as follows: , This indicates the electricity load at time t on the same date last year, as published by the marketing center. If the same date last year was not a holiday, then the electricity load at time t on the last day of the same holiday last year is used. This represents the electricity load at time t on the same date last year in the user-side data collection. If the same date last year was not a holiday, then the user's electricity load at time t on the last day of the same holiday last year is used.
7. The method according to any one of claims 1 to 6, characterized in that, For the second prediction user list, a second prediction method is called; in the second prediction method, according to the context information of the interaction between the user management agent and the administrator, it is confirmed whether the user has production scheduling changes; if there is no change, the first prediction method is executed for the second prediction user list; if there is a change, the user is guided to provide the time of the change and the related information of the electricity load, and the information is combined into system prompt words to generate adjusted electricity load prediction data by the user management agent reasoning.
8. The method according to any one of claims 1 to 6, characterized in that, Skill technology is used to plan the load prediction process to obtain the first prediction user list and the second prediction user list.
9. A system for user electricity load prediction and regulation based on agent interaction, characterized in that Comprising: A data collection module: collecting historical electricity load data of target users or user groups from a marketing operation platform through a timing task; A calculation module: calculating the average daily electricity data of each user in the collected historical electricity load data to obtain user daily average electricity scale data; A storage module: sorting users using the daily average electricity scale data, and saving the users and their daily average electricity scale data in the form of key-value pairs to a database; The user management intelligent agent module is configured to interact with the administrator through natural language, initiate daily user management confirmation information to the administrator, and receive confirmation instructions of the administrator about user information of users requiring special load prediction and data types for load prediction; The list acquisition module is configured to use the confirmed user information and the data types for load prediction as prompt words, plan a load prediction process, and obtain a first predicted user list and a second predicted user list; The prediction module is configured to call corresponding prediction methods for the first predicted user list and the second predicted user list to perform prediction through MCP technology.