DNN-Based Prediction Method of Optimal Load Shedding for Power System With High Penetration of Renewable Energy Under Typhoon Weather

Extreme natural disasters such as typhoons have a serious impact on the power system, and the probability of power equipment failure will increase dramatically, bringing uncertainty that will affect the operation of the power system. And with the increasing penetration of renewable energy sources such as wind power in the power system, the uncertainty and volatility of the power system operation state have increased dramatically. In this paper, we consider optimal load shedding with N-1 faults in the power system as the research object, and in addition to considering branch openings, we also need to consider renewable energy sources and load fluctuations, and the proliferation of the number of scenarios brings about the problem of high computational cost and computational time. To address the above problems, a DNN-based prediction method of optimal load shedding for power system with high penetration of renewable energy under typhoon weather is proposed, which firstly constructs a wind farm scenario under the influence of typhoon, and then establishes a deep neural network-based optimal current model to effectively explore the nonlinear characteristics of nonlinear energy flow equations. Using the trained deep neural network model, the N -1 fault analysis results are obtained by input feature mapping, and the calculation results represented by load curtailment are obtained. Finally, the proposed method is analysed and validated by IEEE case 24 test system. Based on the validation, the proposed deep learning-based optimal curtailment load forecasting method for high percentage renewable energy grids under typhoon weather is able to calculate the power system state quickly and accurately.