Metaheuristic Optimization for N-2 Contingency Analysis in Power Transmission Networks: A Comparative Benchmark Study Using SA and GA

Abstract

N-2 contingency analysis is an integral part of securing the operation of modern power transmission systems. Even though N-1 analysis is typically applied, exhaustive N-2 (dual-line) contingency analysis is computationally infeasible in large scale systems. In this paper, a comparison of the performance of simulated annealing and genetic algorithm is carried out regarding their ability to determine the critical N-2 contingencies in IEEE benchmark test systems such as the 14-bus, 30-bus, 57-bus, and 118-bus systems. Moreover, a novel severity score function consisting of VSA, topological islanding, lines' overload, and cascading failure risk is designed. Furthermore, Subspace Confidence, which measures the correlation between inferred weaknesses according to VSA and combinational risk is suggested as a new metric for the assessment process. The computational analysis reveals that both simulated annealing and genetic algorithm can find N-2 catastrophes (severity =10.0) with a reduction of 97.3% in the number of evaluations with respect to exhaustive analysis in IEEE 118-bus network. More interestingly, although the IEEE 14-bus test network is safe under N-1, N-2 analysis can detect some hidden weaknesses resulting in total blackout in the network.