A Simple Opposition-based Greedy Heuristic Search for Dynamic Economic Thermal Power Dispatch

This article proposes a simple opposition-based greedy heuristic search to solve a dynamic thermal power dispatch problem as a non-linear constrained optimization problem in the constrained search space. Opposition-based learning is applied at two stages. First, an initial population is generated to select good candidates by extensively exploring the search space. Second, it is implemented for migration to maintain diversity in the set of feasible solutions. The proposed method applies a mutation strategy by perturbing the genes heuristically and seeking a better one, which introduces parallelism and makes the algorithm greedy for a better solution. The greediness and randomness pulls the algorithm toward a global solution. Acceleration of the algorithm is independent of any parameter tuning. Feasible solutions are achieved heuristically by modifying the generation schedules within operating generation limits. Opposition-based greedy heuristic search has been implemented to analyze dynamic economic thermal power dispatch problems considering ramp-rate limits, prohibited operating zones, valve-point-loading effects, and transmission losses encountered in realistic power system operation. The validity of the proposed method is demonstrated on medium and large power systems. Opposition-based greedy heuristic search emerges as competitive with existing solution techniques. A Wilcoxon signed-rank test also proves the supremacy of opposition-based greedy heuristic search.