Two-Level Hierarchical Approach to Unit Commitment Using Expert System and Elite PSO

This paper presents a two-level hierarchical approach that combines an expert system (ES) with an elite particle swarm optimization (EPSO) to solve the unit commitment (UC) problem. Existing UC solution methods have the problems of stopping at a local optimum and slow convergence when applied to large-scale, heavily constrained UC applications. In this work, an ES is developed to handle all constraints. The ES is initially used as a pre-dispatch tool to create a robust swarm. Then, the ES and EPSO are combined to seek the optimal solution. All constraints are incorporated and satisfied during the pre-dispatch and the evolution process. Hence, infeasible positions in the solution space will not be visited. The execution time of the proposed approach grows approximately linearly, rather than geometrically, with problem size. This feature is attractive in large-scale systems. The proposed approach is successfully applied to a popular test system of up to 100 units and the real Taipower 40-unit 168-h system. Both solution cost and execution time are superior to those of published methods. Total fuel costs of the proposed approach are several millions of dollars less than those of existing methods in the Taipower case.