一种基于多核相关向量机的模拟电路故障预测方法  

Analog circuit fault prognostic approach based on MKRVM

作　　者：张朝龙 何怡刚[2,3] 杜博伦[2] 张兰芳[1] 江善和 Zhang Chaolong;He Yigang;Du Bolun;Zhang Lanfang;Jiang Shanhe(School of Physics and Electronic Engineering,Anqing Normal University,Anqing 246011,China;School of Electrical Engineering and Automation,Wuhan University,Wuhan 430072,China;School of Electrical Engineering and Automation,Hefei University of Technology,Hefei 230009,China)

机构地区：[1]安庆师范大学物理与电气工程学院,武汉430072 [2]合肥工业大学,安庆246011 [3]武汉大学电气与自动化学院电气与自动化工程学院,合肥230009

出　　处：《电子测量与仪器学报》2019年第11期95-101,共7页Journal of Electronic Measurement and Instrumentation

基　　金：国家自然科学基金(51607004,51577046,51977153);国家自然科学基金重点(51637004);国家重点研发计划“重大科学仪器设备开发”(2016YFF0102200);装备预先研究重点项目(41402040301);安徽高校自然科学研究重点项目(KJ2018A0369)资助

摘　　要：针对模拟电路的稳定是电子设备正常运行的重要保证,提出了一种用于模拟电路故障预测的方法。首先提取模拟电路的输出电压为预测特征,计算测试元件老化过程中与标称值时响应电压的皮尔逊相关系数,作为测试元件的健康度,并以此为故障预测的样本数据。提出了一种多核相关向量机(MKRVM)方法,基于测试元件的健康度数据,对测试元件的故障发生时间开展预测。在该方法中,MKRVM核函数的系数由量子粒子群算法优化生成。最后以四运放双二阶高通滤波器电路的故障预测实验验证了提出的方法的有效性和先进性。The stability of analog circuits is an important guarantee for normal operation of electronic equipments, an analog circuit fault prognostic approach is presented. Firstly, output voltages are extracted as prognostic features. Then, the component’s healthy degrees are generated by calculating Pearson product-moment correlation coefficient(PPMCC) between voltages of component’s aging condition and no-fault condition, and they are used as sample data. The healthy degree data are learnt by the proposed multiple kernel relevance vector machine(MKRVM) algorithm, and a model is constructed to estimate failure time of the component. Meanwhile, the kernels’ weights are generated by a quantum-behaved particle swarm optimization(QPSO) algorithm. Finally, a four-op-amp bi-quad highpass filter circuit fault prognostic experiment reveals the effectiveness and advancement of the proposed approach.

关 键 词：模拟电路 故障预测 健康度 多核相关向量机 量子粒子群 

分 类 号：TH707[机械工程—仪器科学与技术]