修正的差分进化算法在系统可靠性中的应用

Application of modified differential evolution algorithm in system reliability

提出了一种修正的差分进化算法(MDE)以解决4种系统可靠性问题,它们分别为串联系统、复杂(桥)系统、串并联系统和超速保护系统。MDE算法对DE算法的变异操作进行了一种有效的改进,使得所有解向量以递增的概率朝全局最优解方向移动。另外,罚函数法被用来处理目标函数值和资源约束之间的关系。相应地,通过结合MDE算法和罚函数法以获得令人满意的可行解。实验结果表明,MDE算法所获得的4种系统的可靠性都要好于文献中所报道的结果,分别为0.931682、0.999889 63、0.999976 65和0.999955。

A modified differential evolution algorithm（MDE） is proposed to solve four system reliability problems,which are those of series system,complex（bridge） system,series-parallel system and overspeed protection system,respectively.The MDE algorithm has made an efficient improvement on the mutation operation of the DE algorithm,and it enables all solution vectors to move to the global best solution vector with increased probability.In addition,we adopt the penalty function method to deal with the relationship between objective function value and resource constraints.Thus,combining MDE algorithm and penalty function method,we obtain satisfactory feasible solutions for system reliability problems.Experimental results show that the reliabilities of four systems obtained using the MDE algorithm are all better than those previously reported in recent literatures,and they are 0.931682,0.999 889 63,0.999 976 65 and 0.999 955,respectively.