提出了一种克隆选择和聚类的模拟电路故障诊断技术,该技术将克隆选择算法的全局寻优和模糊C-均值的局部寻优相结合,并利用模糊C-均值的目标函数去构造克隆选择算法的亲和力函数,可以减少特征空间中样本的相互重叠所产生的诊断不确定性,利于故障模式的定位。仿真结果和分析验证证明:该方法缩短了故障的分类器收敛时间,提高了故障诊断率。
A new method of analog fault diagnosis technology based on clonal selection and clustering is proposed. It can not only combine global optimizing ability of clonal selection algorithm( CSA )and local optimizing ability fuzzy C-means(FCM) ,but also utilize the target function of the FCM to construct the affinity function of the CSA. It can minimize the diagnosis uncertainty which is caused by overlapped samples and is advantage of fault location. Simulation results and analizing verify the proposed method curtails fault sorting convergence time and improves the fault diagnosis precision.