A Novel Approach for Analog Circuit Fault Prognostics Based on Improved RVM

被引：40

作者：

Zhang, Chaolong ^{[1
,2
]}

He, Yigang ^{[1
]}

Yuan, Lifen ^{[1
]}

Deng, Fangming ^{[1
]}

机构：

[1] Hefei Univ Technol, Sch Elect Engn & Automat, Hefei 230009, Peoples R China

[2] Anqing Normal Univ, Sch Phys & Elect Engn, Anqing 246011, Peoples R China

来源：

JOURNAL OF ELECTRONIC TESTING-THEORY AND APPLICATIONS | 2014年 / 30卷 / 03期

基金：

中国国家自然科学基金;

关键词：

Analog circuits; Frequency features; Cosine distance; Fault prognostics; RVM; PSO; NEURAL-NETWORK APPROACH; WAVELET TRANSFORM; DIAGNOSIS; SELECTION; PSO;

D O I：

10.1007/s10836-014-5454-8

中图分类号：

TM [电工技术]; TN [电子技术、通信技术];

学科分类号：

080906 [电磁信息功能材料与结构]; 082806 [农业信息与电气工程];

摘要：

In order to estimate the remaining useful performance (RUP) of analog circuits precisely in real time, an analog circuit fault prognostics framework is proposed in the paper. Output voltages are extracted from circuit responses as features to calculate cosine distance which can reflect the health condition of analog circuits. Relevance vector machine (RVM) which has been improved by particle swarm optimization (PSO) algorithm is applied to estimate the RUP. Twelve case studies involving bandpass filter, highpass filter and nonlinear circuit have validated the predict performance of the approach. Simulation results demonstrate that the proposed approach has higher prediction precision.

引用

页码：343 / 356

页数：14

共 29 条

[1]

Prognostics of Failures in Embedded Planar Capacitors using Model-Based and Data-Driven Approaches [J].