不确定多属性决策的单目标优化方法研究

运用神经网络的非线性映射和遗传算法的寻优特性,建立了不确定多属性决策的单目标优化模型。把属性值区间作为遗传算法染色体的搜索范围,用训练好的神经网络计算适应度。用不同的适应度函数来计算综合属性值区间数的下界和上界,然后对方案进行排序。算例结果表明,该方法是可行的。

Singular-objective optimization method of uncertain multi-attribute decision-making

The neural network and genetic algorithm were used in the multi-attribute decision-making.When the elements in decision-making matrix were in the forms of interval number,the genetic algorithm was used to search the best elements in the forms of interval number with the maximum or minimum evaluation.The evaluation was produced by the BP neural network which was trained by the information from certain multi-attribute decision-making.The integrated attribute was composed of the minimum and maximum evaluation produced by the genetic algorithm.A possibility degree-based method for ranking alternatives was utilized for ranking alternatives of the integrated attribute,and a numerical example was also used to show the feasibility of the developed method.