Research on Load Intelligent Transfer Strategy of Urban Power Grid Based on Improved DQN

Abstract

With the expansion of the scale of the urban power grid, its structure is becoming more and more complex. After the occurrence of N-1 failure, the operation risk is likely to increase, resulting in a large area of power failure at N-1-1. Therefore, this paper proposes a load transfer strategy of urban power grid based on Double Deep Q-Network to generate load transfer scheme of power grid intelligently. Firstly, the load transfer problem of urban power grid is modeled as a sequential decision problem which is easy to be learned and trained by agents, and the intelligent load transfer decision model is constructed. Secondly, a learning framework of agent load transfer knowledge is proposed by using a large amount of interaction information between agents and line switches in the simulation environment to continuously accumulate power grid operation knowledge. Finally, in order to improve the effectiveness and generalization of the algorithm, a strategy of pre-action-change exploration value selection is added. The effectiveness of the proposed method is verified through the analysis of actual power grid examples. The proposed method can reduce the serious consequences such as the loss of voltage in all substation stations after N-1-1 occurrence, and provide support and guarantee for the operation safety of urban power grid.