Design of Multi-Machine Power System Stabilizer using Evolutionary Algorithm

S. Sheetekela, S.P. Chowdhury, and K.A. Folly (South Africa)


Stability, breeder genetic algorithm, cross over, premature convergence, low frequency oscillations, eigenvalues


This paper presents the design of power system stabilizers using evolutionary algorithms. Three techniques were considered, namely: Population Based Incremental Learning (PBIL), Genetic Algorithm (GA) and the Breeder Genetic Algorithm (BGA) with adaptive mutation. Eigen value analysis was used in the objective for the respective PSS designs, whereby the lowest damped ratio was to be maximized. Simulation was used to compare the performance of the PSSs designed using the different techniques. Theoretically BGA optimizes slightly better than PBIL, while PBIL gives better results than GA. Overall evolution algorithm techniques work better than conventional methods, which is the CPSS. As a verification of the above, simulation results are presented for multiple operating conditions with PSS designed with all the above mentioned methods. Time domain for both smaller and larger disturbances showed that PBIL and BGA performed slightly similar, but performed better than GA. All evolution algorithms perform better than CPSS.

Important Links:

Go Back