摘要
经典多目标决策方法通过直接或利用层次分析法给定各目标权因子,再进行多目标综合。这种算法存在局限性,如缩放因子或模糊集参数实时调整难、获取最优解效率低等。现有舰船电网故障恢复方法仅给出负荷投切方案,且不考虑舰船线损。在严格定义"支配"、"目标支配"和"约束支配"3个概念的基础上,利用解的非支配关系,提出采用本质多目标进化算法并面向开关的故障智能恢复决策,该方法毋需权因子、缩放因子和模糊集参数,并实现了故障恢复后线损最小,在获得Pareto解集后,再作目标偏好选择,给出最优开关动作序列,提高了获取最优解的效率,解决了多约束条件下解的支配关系判别难题。算例分析证实了该方法可行、有效,且优于经典多目标方法。
In the classical multiobjective decision-making method,the multiple objectives are integrated by using their weighting factors given directly or through the analytic hierarchy process（AHP）.There are some limitations in adjusting the scaling factors or fuzzy set parameters in real time and in obtaining the optimal solutions efficiently.In the present service restoration methods of shipboard power network,the load switching scheme is only given and the line losses are not being considered.Based on strict definition of ＇domination＇,＇objective domination＇,and ＇constraint domination＇,utilizing the nondominated relation between solutions,a switch-oriented and intelligent service restoration strategy of shipboard power network using the nature multiobjective evolutionary algorithm was proposed.The strategy did not need the weighting factors,scaling factors,and fuzzy set parameters.And the objective of the minimization of line losses after service restoration was considered.The optimal solution was selected by the preferential relation among objectives after obtaining a Pareto solution set.Also,the optimal switch operation sequence was given.The efficiency of obtaining the optimal solution was improved.The problem of the determination of the dominated relation between solutions under the multiconstraint condition was solved.The analysis results of cases verified that the proposed method was feasible and effective and better than the classical multiobjective approach.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2011年第31期118-124,共7页
Proceedings of the CSEE
基金
国家重点基础研究发展计划项目(973项目)(2009CB219703)~~
关键词
故障恢复
网络重构
多目标进化算法
PARETO前沿
舰船电力网络
service restoration
network reconfiguration
multiobjective evolutionary algorithm
Pareto front
shipboard power network
