|国家科技期刊平台
首页|期刊导航|机械制造与自动化|基于优化SVM的BUCK电路故障诊断方法

基于优化SVM的BUCK电路故障诊断方法OACSTPCD

Fault Diagnosis Method of BUCK Circuit Based on SVM Optimization

中文摘要英文摘要

核心功率器作为功率变换器的重要组成部分,一旦发生故障,直接影响电路的安全运行.为此设计核心功率器件的加速退化实验方案,采用加速退化实验中退化程度最为严重的电解电容和SiC MOSFET功率管代表DC-DC变换器的软故障器件.实验设定5种工况条件,分别采集每种工况条件下的4种电路信号.采用ReliefF算法对48维特征进行特征优选,采用粒子群算法优化支持向量机(PSO-SVM)进行故障分类,并与SVM、KNN分类算法进行对比分析,验证了所提方法的优越性.实验结果表明:PSO-SVM故障诊断方法可以获得更高的故障诊断率.

As an important part of the power converter,the failure of the core power converter will directly affect the safe operation of the circuit.Therefore,this paper designs the accelerated degradation experiment of core power devices,and the electrolytic capacitor and SiC MOSFET power tube with the most serious degradation degree in the accelerated degradation experiment are adopted to represent the soft fault devices of DC-DC converter.Five working conditions are set to collect four circuit signals under each working condition.Relief algorithm is used to optimize the 48-dimensional features,particle swarm algorithm is applied to optimally support vector machine(PSO-SVM)for fault classification,and comparison by SVM and KNN classification algorithm is conducted,which verifies the superiority of the proposed method.The experimental results show that the PSO-SVM fault diagnosis method can obtain higher fault diagnosis rate.

许煜辰;王友仁;常烁

南京航空航天大学 自动化学院,江苏南京 210016

计算机与自动化

功率变换器SiC MOSFET功率管加速退化实验PSO-SVM

power converterSiC MOSFET power tubeaccelerated degradation testPSO-SVM

《机械制造与自动化》 2024 (002)

220-223,273 / 5

南京航空航天大学研究生科研与实践创新计划项目(xcxjh20210329)

10.19344/j.cnki.issn1671-5276.2024.02.046

评论