寻求“理想” 解的改进多目标粒子群优化算法

如何在众多非劣解中为决策者推荐一个合理的方案是使用多目标粒子群算法(MOPSO) 所面临的问题. 为此, 将逼近理想解的排序方法(TOPSIS 策略) 引入到算法中. 为了提高求解精度和均匀性, 还提出了基于Pbest 的变异策略和改进的?? 邻近距离策略. 测试结论显示, 仅使用TOPSIS 策略确定Gbest 的算法, 求解精度虽好, 但均匀性较差, 而包含所有改进策略的算法在精度和均匀性方面都更优, 并且能够按照TOPSIS 方法在非劣解集中找到一个适合向决策者推荐的“理想” 方案.

Improved multi-objective particle swarm optimization algorithm that can give “ ideal ” solution

One problem of using multi-objective particle swarm optimization algorithm(MOPSO) is how to find a recommendable solution within the non-inferior solution set for decision-makers. Therefore, the technique for order preference by similarity to ideal solution(TOPSIS) strategy is introduced into the MOPSO algorithm. In order to improve the accuracy and uniformity of solutions, two other strategies are also proposed: the mutation operator according to Pbest, and the modified method to calculate ?? neighbor distance. The results show that the solution uniformity of the algorithm which chooses Gbest using the TOPSIS method only is not satisfied, while the algorithm that uses all improvement strategies has better performance. An ideal solution can be found for decision-makers according to the TOPSIS strategy.