A Maiden Application of Competitive Swarm Optimizer for Solution of Economic Load Dispatch with Parameter Estimation

Abstract

This work, presents a novel application of newly developed Competitive Swarm Optimizer (CSO) in solving a non-linear, non-convex and constrained Economic Load Dispatch (ELD) problem of power systems. A comparative analysis is performed between Particle Swarm Optimization (PSO) and CSO. CSO is basically inspired by the behavior of well-known PSO algorithm. Similar to PSO, CSO is also swarm based optimization technique and has both cognitive as well as social component. The difference lies in the fact that CSO neither uses local best nor global best in updating the position of their particles, which makes the algorithm memory free. A pair wise competition is performed and only loser particles are updated after getting experience from the winner one. The performance of algorithm is tested on 4 benchmark systems with 5 case studies. Case studies include both small as well as bigger system with various degree of constraints such as; Power balance, Prohibited Operating Zone (POZ), ramp limits, valve point loading etc. Moreover, general and standard statistical tests (t-test) are also performed to investigate the consistency and robust of the proposed algorithm. The performance of CSO is significantly dependent on its social factor (ϕ) and population size. In view of this, in the present work the performance of CSO with the variation in population size and (ϕ) are also studied. Parametric study is also performed for all cases to judge the sensitivity of the algorithm. The study reveals that the proper tuning of social factor (ϕ) and population size significantly reduce the search space which helps the algorithm to accelerate its convergence.