A new intelligence solution for power system economic load dispatch

Abstract

Economic Load Dispatch is one of the most important tasks to be performed in the operation and planning of a power system that decides the generation schedule of generating units with an objective of minimizing the total fuel cost. Normally, the fuel cost of generators can be treated as a quadratic function of real power generation. In fact, valve point loading effect in thermal power plants introduces discontinuity. The classical optimization methods require continuous differentiable objective functions; therefore they at times provide global minima. The evolutionary computation methods can handle non-differential and non-convex objective functions and provide global or near global optimum solutions. Evolutionary techniques such as Genetic Algorithm (GA), Differential Evolution (DE), and Particle Swarm Optimization (PSO) saw wide applications in economic load dispatch. Similar to evolutionary computation, physical behaviour intelligence called gravitational search intelligence is recently developed and has not been applied in many fields. Use of gravitational search intelligence not only avoids coding and monotonous decoding as prevalent in transformations of GA but also results in less burden on parameter settings, population size, number of iterations and no memory requirement of solution as in PSO. In this paper, Gravitational Search algorithm (GSA) is applied to economic load dispatch problem with valve point loading and Kron's loss. Its performance is compared for accuracy and speed with contemporaries heuristic search techniques like PSO, DE, and GA and traditional method sequential quadratic programming (SQP) on 3, 6, 13 and 40-unit test systems. The simulation results reveal that GSA has a great potential in handling complex optimization problems and capable to discover quality solution quickly even for large scale systems.