基于二次寻优和可行域坍缩的梯级水库调度研究OA

To address the conflict between solution efficiency and accuracy caused by the curse of dimensionality and the strong coupling of multiple constraints in large-scale cascade reservoir scheduling,a two-stage hybrid solution framework based on secondary optimization and adaptive feasible region collapsing was proposed in this paper.First,the improved particle swarm optimization algorithm and the improved genetic algorithm were adopted to conduct a parallel global search,and a diversified high-quality initial solution group was rapidly generated by utilizing a flexible constraint mechanism.Second,an adaptive feasible region collapsing mechanism based on constraint state awareness was established.For the compliant solution set,the envelope was extracted and the safety margin was expanded to construct the collapsed search domain,while for the violation solution set,the full space was maintained to avoid being misled.Finally,taking the optimal initial solution as the starting point,the secondary optimization was carried out within the collapsed feasible region by utilizing the improving progressive optimality algorithm(SAAPOA),and the strict feasibility of the solution was ensured through an adaptive annealing step size and an online correction mechanism.The case study of the joint operation of six reservoirs in the main stream of the Yangtze River shows that the power generation benefits of this method at both ten-day and daily scales are superior to those of single algorithms,the computation time is significantly reduced compared with the SAAPOA,and the synergy between global extensive search and local refinement is effectively realized.