Integration of Parallel EPSO and Variable TS for Unit Commitment with Nonsmooth Fuel Cost Functions

In this paper, a new hybrid meta-heuristic method is proposed to solve a unit commitment (UC) problem with nonsmooth fuel cost functions effectively. The proposed method focuses on global optimization in a sense that a generation company need carry out the cost reduction under competitive environment. The proposed method integrates parallel evolutionary particle swarm optimization (PEPSO) with variable neighborhood tabu search (VTS). The objective of UC is to minimize operation-cost while satisfying the constraints. The unit commitment problem is hard to solve due to the complexity in determining on-off conditions and output of generators. The problem formulation may be written as a nonlinear mixedinteger problem. In addition, large steam turbine generators with nonsmooth fuel cost functions are considered from a realistic standpoint. This paper proposes a new hybrid meta-heuristic method that combines VTS with PEPSO and evaluates solutions with two layers. Layer 1 determines the on-off conditions of generators with VTS while Layer 2 evaluates output of generators with PEPSO. TS is very effective for solving a combinatorial optimization problem efficiently. EPSO has better performance in dealing with an optimization problem of continuous variables. However, both methods still have room to improve solution quality and reduce computational time. Therefore, TS is improved to include the technique of the priority list limit and variable neighborhood search and EPSO is enhanced by the parallel scheme with the island model. The effectiveness of the proposed method is successfully applied to sample systems.