CN-122026312-A - Novel power management method and device based on large model and electronic equipment

CN122026312ACN 122026312 ACN122026312 ACN 122026312ACN-122026312-A

Abstract

The invention discloses a novel power management method and device based on a large model and electronic equipment, wherein the method comprises the following steps: and acquiring power management instructions, power data, industry data, holiday data, social event data and socioeconomic data, reasoning the data by utilizing a large model to obtain source load information and power scheduling information, and continuously fusing the source load information and the power scheduling information to obtain a power scheduling strategy, and controlling corresponding intelligent agents to execute the power scheduling strategy. According to the technical scheme, multiple complex power prediction and scheduling schemes can be integrated into an intelligent power system through a unified large model, and 'considerable, measurable, controllable and adjustable' closed-loop management is realized.

Inventors

Han Huhu

Assignees

国能内蒙古电力蒙西新能源有限公司

Dates

Publication Date: 20260512
Application Date: 20251229

Claims (10)

1. A novel power management method based on a large model, the method comprising: acquiring power management instructions, power data, industry data, holiday data, social event data and socioeconomic data; The power data, the industry data, the holiday data, the social event data and the socioeconomic data are sent into a large model, wherein the large model comprises a base model, a source prediction model, a load prediction model and a scheduling model, and the base model comprises a data marking layer, a data distribution layer and a knowledge learning layer; Marking the electric power data, the industry data, the holiday data, the social event data and the socioeconomic data by using the data marking layer respectively, and mapping the marked data and the marked data into high-dimensional vectors; Splitting the power management instruction into a plurality of subtasks by using the base model, and determining the execution sequence among the subtasks; The data distribution layer sends the power data sub-vector in the high-dimensional vector to the source prediction model to obtain power supply information according to the sub-tasks, the execution sequence among the sub-tasks, the power knowledge learned by the knowledge learning layer and the marks; The data distribution layer sends the electric power data sub-vector, the industry data sub-vector, the holiday data sub-vector and the social event sub-vector into the load prediction model according to the subtasks, the execution sequence among the subtasks, the electric power knowledge learned by the knowledge learning layer and the marks to obtain load information; the data distribution layer sends the socioeconomic data subvector, the power supply information and the load information to the scheduling model according to the subtasks, the execution sequence among the subtasks, the power knowledge learned by the knowledge learning layer and the marks to obtain power scheduling information; Fusing the power management instruction, the subtasks, the execution sequence among the subtasks, the power supply information, the load information and the power scheduling information to obtain a power management strategy, wherein the power management strategy comprises strategy contents and intelligent agents corresponding to the strategy contents; And controlling the corresponding intelligent agent to execute the action corresponding to the strategy content according to the power management strategy.
2. The method of claim 1, wherein the power data includes electrical quantity data, meteorological data, geographical information of a location of each power device, historical power generation data, historical power consumption data, and device parameters of each power device.
3. The large model-based novel power management method of claim 1, wherein the base model is deployed at a cloud end, and the source prediction model, the load prediction model and the scheduling model are deployed at different edge nodes, respectively.
4. The large model-based novel power management method of claim 1, wherein the large model further comprises a fault detection model; the data distribution layer sends the socioeconomic data subvector, the power supply information and the load information into the scheduling model according to the subtasks, the execution sequence among the subtasks, the power knowledge learned by the knowledge learning layer and the marks, and the data distribution layer further comprises the following steps before obtaining the power scheduling information: Acquiring inspection data of each terminal aiming at an agent at the terminal; The data distribution layer sends the power data subvector and the inspection data to the fault detection model to obtain equipment fault information according to the subtasks, the execution sequence among the subtasks, the power knowledge learned by the knowledge learning layer and the marks; The data distribution layer sends the socioeconomic data subvector, the power supply information and the load information to the scheduling model according to the subtasks, the execution sequence among the subtasks, the power knowledge learned by the knowledge learning layer and the marks, and the power scheduling information comprises: the data distribution layer sends the power scheduling information into the scheduling model according to the subtasks, the execution sequence among the subtasks, the power knowledge learned by the knowledge learning layer, the marks, the power supply information, the load information and the equipment fault information to obtain the power scheduling information.
5. The large model-based novel power management method of claim 1, wherein the large model further comprises a risk assessment model; the data distribution layer sends the socioeconomic data subvector, the power supply information and the load information into the scheduling model according to the subtasks, the execution sequence among the subtasks, the power knowledge learned by the knowledge learning layer and the marks, and the data distribution layer further comprises the following steps before obtaining the power scheduling information: Acquiring construction process data, land planning permission data, critique data and construction permission data of each terminal; the data distribution layer sends the construction process data, the land planning permission data, the criticizing data, the construction permission data and the power data sub-vectors into the risk assessment model according to the subtasks, the execution sequence among the subtasks, the power knowledge learned by the knowledge learning layer and the marks to obtain construction safety risk information; correspondingly, the data distribution layer sends the socioeconomic data subvector, the power supply information and the load information to the scheduling model according to the subtasks, the execution sequence among the subtasks, the power knowledge learned by the knowledge learning layer and the marks, and the power scheduling information comprises: The data distribution layer sends the power scheduling information into the scheduling model according to the subtasks, the execution sequence among the subtasks, the power knowledge learned by the knowledge learning layer, the marks, the power information, the load information and the construction safety risk information, and the power scheduling information is obtained.
6. The method of claim 1, wherein the fusing the power management instruction, the subtasks, the execution sequence among the subtasks, the power supply information, the load information, and the power scheduling information to obtain a power management policy comprises: And fusing the power management instruction, the subtasks, the execution sequence among the subtasks, the power supply information, the load information and the power scheduling information to obtain an intermediate result, and determining a power management strategy according to the intermediate result, the power grid peak regulation capacity and the power consumption plan.
7. A new power management device based on a large model, the device comprising: the acquisition module is used for acquiring power management instructions, power data, industry data, holiday data, social event data and socioeconomic data; the system comprises a power data acquisition module, a power management module and a power management module, wherein the power data acquisition module is used for acquiring power data, industry data, holiday data, social event data and socioeconomic data, the power data, the business data, the holiday data, the social event data and the socioeconomic data, the business data are transmitted to a large model, the large model comprises a base model, a source prediction model, a load prediction model and a scheduling model, the base model comprises a data marking layer, a data distribution layer and a knowledge learning layer, the large model also corresponds to a plurality of intelligent bodies, the intelligent bodies are deployed at a terminal, and the terminal comprises power equipment and energy storage equipment; The mapping module is used for respectively marking the electric power data, the industry data, the holiday data, the social event data and the socioeconomic data by utilizing the data marking layer and mapping the marked data and the marked data into high-dimensional vectors; The splitting module is used for splitting the power management instruction into a plurality of subtasks by utilizing the base model and determining the execution sequence among the subtasks; The power information acquisition module is used for the data distribution layer to send the power data sub-vector in the high-dimensional vector into the source prediction model to obtain power information according to the sub-tasks, the execution sequence among the sub-tasks, the power knowledge learned by the knowledge learning layer and the mark; The load information acquisition module is used for the data distribution layer to send the electric power data sub-vector, the industry data sub-vector, the holiday data sub-vector and the social event sub-vector into the load prediction model according to the subtasks, the execution sequence among the subtasks, the electric power knowledge learned by the knowledge learning layer and the marks to obtain load information; The scheduling information acquisition module is used for the data distribution layer to send the socioeconomic data subvectors, the power supply information and the load information into the scheduling model according to the subtasks, the execution sequence among the subtasks, the power knowledge learned by the knowledge learning layer and the marks to obtain power scheduling information; The result fusion module is used for fusing the power management instruction, the subtasks, the execution sequence among the subtasks, the power supply information, the load information and the power scheduling information to obtain a power management strategy, wherein the power management strategy comprises strategy contents and intelligent agents corresponding to the strategy contents; and the policy execution module is used for controlling the corresponding intelligent agent to execute the action corresponding to the policy content according to the power management policy.
8. A computer program product comprising computer programs/instructions which, when executed by a processor, implement the steps of the method of any of claims 1-6.
9. An electronic device comprising at least one memory for storing a computer program and at least one processor that runs the computer program to cause the electronic device to perform the novel large model-based power management method of any one of claims 1-6.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored therein a computer program for use in the electronic device of claim 9.

Description

Novel power management method and device based on large model and electronic equipment Technical Field The invention relates to the technical field of novel power management, in particular to a novel power management method and device based on a large model and electronic equipment. Background Under the general trend of low carbon of global energy, the permeability of renewable distributed power generation, distributed energy storage, flexible load and other distributed resources mainly based on wind and light in a power distribution network is increased year by year. However, as an important physical carrier and a foundation stone of the novel power system, the uncertainty, openness and complexity increase of numerous adjustable resources in the novel power system bring great challenges to the integration of source network charge storage and wind-solar energy absorption, and the flexibility and intelligence requirements of the novel power system are difficult to meet. Disclosure of Invention In view of the above problems, an object of an embodiment of the present invention is to provide a novel power management method, apparatus and electronic device based on a large model, so as to solve the defects in the prior art and improve flexibility and intelligence of a power system. According to an embodiment of the present invention, there is provided a novel power management method based on a large model, the method including: acquiring power management instructions, power data, industry data, holiday data, social event data and socioeconomic data; The power data, the industry data, the holiday data, the social event data and the socioeconomic data are sent into a large model, wherein the large model comprises a base model, a source prediction model, a load prediction model and a scheduling model, and the base model comprises a data marking layer, a data distribution layer and a knowledge learning layer; Marking the electric power data, the industry data, the holiday data, the social event data and the socioeconomic data by using the data marking layer respectively, and mapping the marked data and the marked data into high-dimensional vectors; Splitting the power management instruction into a plurality of subtasks by using the base model, and determining the execution sequence among the subtasks; The data distribution layer sends the power data sub-vector in the high-dimensional vector to the source prediction model to obtain power supply information according to the sub-tasks, the execution sequence among the sub-tasks, the power knowledge learned by the knowledge learning layer and the marks; The data distribution layer sends the electric power data sub-vector, the industry data sub-vector, the holiday data sub-vector and the social event sub-vector into the load prediction model according to the subtasks, the execution sequence among the subtasks, the electric power knowledge learned by the knowledge learning layer and the marks to obtain load information; the data distribution layer sends the socioeconomic data subvector, the power supply information and the load information to the scheduling model according to the subtasks, the execution sequence among the subtasks, the power knowledge learned by the knowledge learning layer and the marks to obtain power scheduling information; Fusing the power management instruction, the subtasks, the execution sequence among the subtasks, the power supply information, the load information and the power scheduling information to obtain a power management strategy, wherein the power management strategy comprises strategy contents and intelligent agents corresponding to the strategy contents; And controlling the corresponding intelligent agent to execute the action corresponding to the strategy content according to the power management strategy. In the above-described novel power management method based on the large model, the power data includes electrical quantity data, meteorological data, geographical information of a location where each power device is located, historical power generation data, historical power consumption data, and device parameters of each power device. In the above novel power management method based on the large model, the base model is deployed at a cloud end, and the source prediction model, the load prediction model and the scheduling model are respectively deployed at different edge nodes. In the above-described novel power management method based on a large model, the large model further includes a fault detection model; the data distribution layer sends the socioeconomic data subvector, the power supply information and the load information into the scheduling model according to the subtasks, the execution sequence among the subtasks, the power knowledge learned by the knowledge learning layer and the marks, and the data distribution layer further comprises the following steps before obtaining the power scheduling information: Acquiring inspec