Study of genetic algorithm performance through design of multi-step LC compensator for time-varying nonlinear loads

• Performance of GA is related to processing time and number of generations required for convergence and the convergence itself.
• GA is affected by choosing its parameters and implementation techniques through designing the multi-step LC.
• The performance of GA is widely affected by choosing its parameters and implementation techniques.

Genetic algorithm (GA) is a search mechanism simulating the natural selection and population genetics. The performance of GA is related to processing time and the number of generations required for convergence and the convergence itself. This article studies how the performance of GA is affected by choosing its parameters and implementation techniques through designing the multi-step LC based on performance criteria; maximizing the power factor (PFPF), minimizing the transmission loss (TLTL), or minimizing the voltage total harmonic distortion (VTHDVTHD). The multi-step LC compensator consists of switchable units thus assuming that a single unit is not sufficient to ensure satisfactory results. GA is used to estimate that steps while holding the performance quantities at the corresponding desired values and constraining the compensator values which would create resonance. The contribution of the proposed procedure is demonstrated in examples taken from previous publications. Finally, simulated results show the performance of GA is widely affected by choosing its parameters and implementation techniques and hence it could be improved.

Figure optionsDownload as PowerPoint slideProposed GA based techniques for multi-step LC compensator