Stochastic Optimization Techniques for Economic Dispatch with Combined Cycle Units

Electricity industries worldwide are undergoing a period of profound upheaval. Conventional vertically integrated mechanism is replaced by a competitive market environment. Generation companies have incentives to produce more electricity at lower cost by applying novel technology: combined cycle, integrated gasification combined cycle, fuel switching/blending, and dual boiler etc. Economic dispatch becomes a non-convex optimization problem, which is difficult, even impossible to solve by conventional methods. Genetic algorithm, evolutionary programming, and particle swarm share a common mechanism, stochastic searching per generation. The stochastic property makes evolutionary algorithms robust and adaptive enough to solve non-convex optimization problem. This paper implements GA, EP, PS algorithms for economic dispatch including combined cycle units, and makes a comparison with classical mixed integer linear programming. The trajectory and searching path of each stochastic optimization technique are shown and compared. The numerical results show that the stochastic optimization techniques are capable of providing approximate global optimal solution for non-convex optimization problem