Optimization of power system emergency load shedding with surrogate-assisted proliferation-reduction differential evolution algorithm

• A universal SAPR evolutionary framework for power system emergency load shedding optimization is proposed. • A proliferation strategy of differential evolution is designed to improve the solution quality. • A reduction strategy of differential evolution is designed to improve the optimization speed. Emergency load shedding (ELS) is vital for maintaining the stability and security of power systems following severe power deficiency. To adapt to varying operating conditions induced by the integration of renewable energy, the ELS schemes need to be optimized rapidly and updated online. The widely used ELS optimization methods rely on time-domain simulation, which is time-consuming and struggles to meet the efficiency of online applications. To address the above issues, this paper proposes a surrogate-assisted proliferation-reduction differential evolution algorithm for ELS optimization, aiming to obtain a high-quality and feasible ELS scheme quickly. Firstly, an ELS optimization model considering multiple security constraints is constructed, and a surrogate-assisted proliferation-reduction evolutionary framework is designed to obtain ELS schemes. Then, a proliferation strategy based on differential evolution is proposed to generate many temporary solutions in each generation through diverse mutation operators. It can broaden the search space and improve the solution quality. Finally, a reduction strategy is proposed to prescreen temporary solutions by a data-driven surrogate model. It can significantly improve the optimization speed. The prescreened solutions are validated with time-domain simulation to guarantee the feasibility of ELS schemes. Test results on New England 39-bus system and a provincial system of China demonstrate that the proposed algorithm can achieve a high-quality ELS scheme in a short time.