Wiener核的快速提取算法  

A rapid extraction algorithm of Wiener kernel

作　　者：刘云峰[1] 林海军[1] 朴伟英[1] 曹亚鹏[1] 

机构地区：[1]哈尔滨理工大学测控技术与通信工程学院,哈尔滨150080

出　　处：《电测与仪表》2015年第24期14-18,共5页Electrical Measurement & Instrumentation

基　　金：国家自然科学基金资助项目(61179024)

摘　　要：对于非线性特性未知的模拟电路,可以运用Wiener泛函级数来描述,且Wiener核值可以表示电路特征。但是随着阶数及采样点数的增加,Wiener核值提取运算次数呈指数形式增长,这给应用带来困难。针对这一问题,提出了一种折叠递推快速算法,经分析计算本快速算法比常规算法节省乘法计算量90%以上。在此基础上,运用快速算法与常规算法针对典型电路进行了对比实验,提取的核相同,可以快捷有效的诊断电路的正常状态及不同软故障状态。The Wiener Functional Series are used in describing the analog circuit of an undiscovered nonlinear characteristic and the circuit character is represented as the value of Wiener kernel. However,the complexity of the value of Wiener kernel extraction is growing at an exponential rate with the increases of orders and sampling points and it is difficult for application. In view of this problem,a folded and recursive fast algorithm was proposed in this paper,and over 90% of the computational cost can be reduced with the fast algorithm than primary algorithm based on the careful analysis and calculation. On this basis,the experiments were conducted that compared the Wiener kernel of primary algorithm and fast algorithm applying on typical circuit which was identical and the conditions of circuit among normal and different soft faults were rapidly and efficiently diagnosed.

关 键 词：Wiener核 特征提取 软故障诊断 

分 类 号：TM933[电气工程—电力电子与电力传动]