Operation scheduling method, system and device of power system and medium

Abstract

The invention discloses an operation scheduling method, system and device of a power system and a medium. The method comprises the steps of obtaining operation mode data of the power system; classifying the operation mode data by adopting the trained decision tree classifier to obtain a typical operation scene type; acquiring an optimal scheduling model corresponding to the typical operation scene type, and controlling operation scheduling of the power system by adopting the optimal scheduling model; wherein the typical operation scene type is obtained through clustering analysis, and the optimal scheduling model comprises a deep reinforcement learning model and a novel power system model. According to the method, a plurality of typical operation scenes are divided in a clustering mode, and corresponding optimization scheduling models are called for different typical operation scenes to control scheduling, that is, a deep reinforcement learning optimization scheduling scheme is designed for different typical scenes, so that the performance of a scheduling decision scheme is effectively improved. The method can be widely applied to the field of power systems.

Description

technical field [0001] The present invention relates to the field of power systems, in particular to an operation scheduling method, system, device and medium of a power system. Background technique [0002] Power system operation mode refers to the technical scheme of power system production and operation, which is divided into different types from the operating status (such as normal, maintenance, accident) and time granularity (such as year, month, day, time). The changes in the operation mode of the power system under the new trend are specifically reflected in the changes in the operation rules of the relevant electrical quantities after seasonal loads, new energy sources, and power flows. The data will be high-dimensional and non-linear. The analysis and utilization of the power system operation mode Its analysis results guide that operation scheduling and so on are nonlinear time coupling problems in high dimensions. [0003] For the application of data-driven method...

AI Technical Summary

Problems solved by technology

However, traditional deep reinforcement learning algorithms are increasingly difficult to adapt to the random multi-scenario operation scheduling problem of new power systems. At present, there is no idea of ​​front-end optimization of deep reinforcement learning algorithms by means of scene classification and pattern recognition, and there is no use of data. Drive technology analysis of power system operation mode to improve the research results of dispatch optimization

Images