Novel Constraint Stratification Ranking Methodology Integrated on Dynamic Balance Imperialist Competition Algorithm for Solving the Multi-Objective Optimal Reactive Power Dispatch Problem

摘要

Computer technology provides new possibilitiesfor handling the multi-objective optimal reactive powerdispatch (MOORPD) problems with high-dimension andnon-differentiability. In this paper, a novel constraintstratification ranking methodology integrated on dynamicbalance imperialist competition algorithm (ICA) is proposedto deal with MOORPD problem, and this methodology iscalled NCSR-DBICA. The proposed methodology includesthree sub-methods to solve three typical problems encounteredwhen dealing with MOORPD problems with ICA. Thesetypical problems include the original ICA which is easy to fallinto local optimum, it is difficult to handle the constraints inMOORPD problem and there is no effective solution setranking method. Thus, the dynamic balance strategy (DBS) isproposed to improve the searching ability of ICA, theconstraint-based country stratification mechanism (CCSM) isproposed to deal with the constraint problem, and a novelranking method (NRM) is proposed to solve the rankingproblem. To verify the effectiveness of the improved method,the NCSR-DBICA, MOICA-FS, NSGA-III, NSGA-II andMOPSO-CD were tested on three test systems. Simulationresults show that NCSR-DBICA can find better results,especially in large scale systems. In addition, two indicators:Generational Distance (GD) and Hyper-volume (HV), wereselected to evaluate the diversity, stability and convergence ofthe above algorithms. The evaluation results also verify thesuperiority of NCSR-DBICA.