基于MDPA算法的火电厂多目标负荷优化分配模型

MDPA algorithm based multi-objective load dispatch

在传统的火电厂经济负荷分配模型基础上,综合考虑全厂供电煤耗率、污染物排放量以及全厂负荷升、降时间3个目标,构建了厂级负荷优化分配的多目标模型。将差分粒子群混合算法发展为一种新型的多目标进化(MDPA)算法,即利用擂台赛法和凝聚层次聚类分析方法分别构造和修剪非支配集,同时加入精英保留策略,保留进化过程中的极值点。将该算法应用于以经济、环保、快速3个目标为多目标的厂级负荷优化分配,并与基于非支配排序的多目标优化(NSGA-Ⅱ)算法进行对比。结果表明,MDPA算法较NSGA-Ⅱ算法收敛速度更快,解集分布更均匀。

On the basis of the conventional single-obj ect model only considering the energy consumption,a multi-obj ective optimization model for load dispatch was established,concerning the energy consumption, environmental protection and load response rate.A hybrid multi-obj ective differential evolution （DE）and particle swarm optimization （PSO）algorithm called MDPA was proposed,which uses arena＇s principle to establish non-dominate set,cluster analysis to change the size of no-dominate set,elitist strategy to retain extreme points during evolution.Results in three multi-obj ective load dispatch problems show faster con-vergence rate,higher convergence precision and more even distribution of solution set compared with NS-GA-Ⅱ.