基于区间离散解产生器的r决策变量区间离散多目标优化方法

摘要

针对数学特性要求严格且现有约束处理方法无法有效解决的决策空间离散分布多目标优化问题,提出一种求解决策变量离散分布的多目标进化算法.首先设计一个满足区间离散解的个体产生器,指引种群的搜索方向;然后针对常规约束违反度计算方法无法适用于决策变量离散约束的情况,提出了一种区间离散约束违反度的计算方法,并将满足决策变量离散约束的个体使用改进可行性规则来处理常规的等式约束和不等式约束;最后对考虑决策变量区间离散模型和未考虑决策变量区间离散模型进行仿真比较.结果表明,在相同的仿真条件下,使用本文处理决策变量离散分布的算法,直接在有定义的变量区间循环产生个体,提高了算法的效率,并且满足可行域的非支配个体明显比未考虑区间离散约束的个体数量多,分布均匀性好.%The strict mathematical algorithm and the existing constraint processing method can not be used in the discrete domain of the decision space for the optimization .It proposes a multi -objective evolutionary algorithm for solving the discrete distribution of decision variables .First, it designs an individual generator that satisfies the interval discretization to guide the search direction of the population .Aiming at the case that the conventional calculation method of constraint violation degree can not apply the interval discrete constraint , it proposes a method of calculating interval discrete constraint .And then it uses the improved feasibility rule to deal with con-ventional equality constraints and inequality constraints .The experimental results show that under the same simu-lation conditions , the method of dealing with the discrete distribution constraints of the decision variables is used to generate the individuals directly in the defined inter variable space , the efficiency of the algorithm is im-proved.Therefore, the non-dominated individuals satisfy the feasible domain of the interval discrete constraint algorithm, and are obviously better than those of the discrete constraints .