Approximating low cost state space areas in economic load dispatch with valve-point loading effects

Economic Load Dispatch (ELD) with non-smooth, non-convex cost functions (NSNC CF) has been broadly researched for several decades. Due to the characteristics of the state space it is often tackled using meta-heuristic optimization techniques. The main focus of research to improve the optimization quality and performance has been trying numerous types and variations of meta-heuristics. However has little attention been given to the characteristics of both CF and state space. This paper analyses a six generating unit (GU) test system to derive three assumptions targeting the location of high quality states within NSNC ELD state space. Subsequently these assumptions are implemented into algorithms to improve the performance of meta-heuristics on NSNC ELD-problems. Finally numerical simulation are used to underpin or refute these assumptions.