OPTIMIZATION OF LARGE-SCALE ECONOMIC DISPATCH WITH VALVE-POINT EFFECTS USING A MODIFIED HYBRID PSO-DSM APPROACH

Abstract

This paper proposes a modified hybrid particle swarm optimization (PSO) and the direct search method (DSM) for the solution of large-scale non-convex economic dispatch (NED) problem with valve-point effects. A novel diversity based particle swarm optimization (DPSO) with a fewer iterations required is developed to increase the possibility of generating high quality initial solutions for the DSM. The enhanced direct search method (EDSM) incorporates the parallel nature of evaluation programming into the direct search procedure to enhance its search capacity about global exploration and local optimization using the answer from DPSO as starting points. Many inequality and equality constraints can be handled properly in the direct search procedure. Appropriate setting of control parameters of the proposed hybrid DPSO-EDSM algorithm is also recommended to increase the possibility of occurrence of escaping from local optimal solution. Numerical experiments are included to demonstrate that the proposed hybrid approach can obtain a higher quality solution with better performance than many existing techniques for the large-scale NED application.