A hybrid multi-objective evolutionary algorithm for economic-environmental generation scheduling

Abstract

In this study, a hybrid Multi-Objective Evolutionary Algorithm (MOEA) is proposed for the Multi-Objective Short-Term Unit Commitment (MO-STUC) problem, considering the minimization of operation cost and emissions as objectives. The proposed algorithm is based on a real-coded Differential Evolution (DE) and a two-step function to simultaneously deal with both the scheduling of the state of the units and the dispatching of the power among the committed units. Moreover, a local search technique, which combines two distinct local search procedures based on Pareto dominance and scalar fitness function, is hybridized with the proposed MOEA. In addition, a heuristic repair mechanism and a problem-specific mutation operator are adopted to enhance the algorithm's performance. The method is tested on two frequently studied test systems comprising 10 and 20 units, respectively. The non-dominated fronts provided by the proposed method adequately approximate the Pareto fronts of the test instances. Simulation results reveal that the proposed algorithm outperforms two standard MOEAs, based on DE and Genetic Algorithm, with respect to the employed quality indicators. Furthermore, the beneficial impact of combining the two local search procedures is demonstrated.