Power ystem controller design: A comparison between breeder genetic algorithm and Population Based Incremental Learning

Abstract

This paper discusses the design of Power System Stabilizers (PSSs) using an Adaptive Mutation Breeder Genetic Algorithm (BGA) and Population Based Incremental Learning (PBIL). BGA is a new form of evolutionary algorithm. It uses the same idea of survival of the fittest like the Genetic Algorithms, however unlike GA; BGA uses the concept of artificial breeding, whereby the offspring takes the best characteristics from the parents. PBIL is an abstraction of genetic algorithm, which explicitly maintains the key components contained in GA's population, but abstracts away the crossover operator and redefines the role of population. The paper compares the performance and effectiveness of the PSSs in damping the electromechanical modes. In evaluating the different methods, an eigenvalue based objective function was used in the design of the PSSs whereby the algorithm maximizes the lowest damping ratio over specified operating conditions. Eigenvalue analysis and time domain simulations show that the systems equipped with BGA-PSS and PBIL - PSS perform very closely. It is also shown that BGA and PBIL based PSSs perform better that the Conventional PSS (CPSS) at all the operating conditions considered except at the nominal operating condition where the CPSS was tuned.