Particle Swarm Optimization Based Optimal Reliability Design of Composite Electric Power System Using Non-sequential Monte Carlo Sampling and Generalized Regression Neural Network

This paper presents a new approach based on particle swarm optimization (PSO) for determining the optimal reliability parameters of composite system using non-sequential Monte Carlo Simulation (MCS) and Generalized Regression Neural Network (GRNN). The cost-benefit based design model has been formulated as an optimization problem of minimizing system interruption cost and component investment cost. Solution of this design model requires the analysis of several reliability levels which needs to evaluate EDNS index for those levels. Evaluation of EDNS in non-sequential MCS requires state adequacy analysis for several thousands of sampled states. In conventional approaches, a dc load flow based load curtailment minimization model is solved for analyzing the adequacy of each sampled state which requires large computational resources. This paper reduces the computational burden by applying GRNN for state adequacy analysis of the sampled states. The effectiveness of the proposed methodology is tested on the IEEE 14-bus system.