Optimization of Emergency Load-Shedding Based on Surrogate-Assisted Differential Evolution

Abstract

Emergency load shedding (ELS) is an essential measure to keep system stability. Due to the high cost of load shedding, minimizing the amount is always the goal while satisfying security requirements. The optimization problem is highly nonlinear and can be solved with heuristic algorithms by generating a large number of candidates. However, it is time-consuming to check the feasibility of each candidate by numerical simulation. To address this issue, this paper presents an accelerated ELS optimization method based on surrogate-assisted differential evolution (SADE). The optimization process is driven by differential evolution (DE). Radial basis function (RBF) neural network is adopted as the surrogate model to replace the numerical simulation for checking the security constraints. Only the most promising candidates pre-screened by RBF are evaluated by numerical simulation. The validity of the proposed ELS optimization method is verified with a provincial power system.