[交通会议论文数据库]
WU Yang ZHANG Zhiyong YUAN Jianhua MA Qing
摘要: As a popular approach to solve Multiobjective Optimization Problem (MOP),weighted-sum (WS) method obtains a series of weight-dependent Pareto Optimalities (i.e.multi-objective global optimums) forming Pateto Front.Each priori (preset) combination of single-objective (SO) weights stands for a certain way to compromise all of SO, e.g.a popular opinion is "Balanced weights lead to the equilibrium solution".To verify this notion, this paper proposes a method to obtain adaptive posteriori weights derived from heuristic search rather than human-judged priori weights, so as to generate an unique Equilibrium Pateto Optimality (Equi-PO) out of the Pareto Front of multiobjective-function (MOFunc), where mutual interest of every single-objective-function (SOFunc) is achieved to a certain "equal" extent.The numerical example reveal that an unique Equi-PO is obtainable with adaptive weights converging towards an unique end, and:(1) For and only for the WS-MOP whose Pareto Front is symmetric to the Equiangular Utopia Ray, "balanced weights" results in "equilibrium solution";(2) For other conditions,"balanced weights" can't.