基于GMKL-SVM的模拟电路故障诊断方法

Analog circuit fault diagnosis based on GMKL-SVM method

提出了一种新颖的基于广义多核支持向量机(GMKL-SVM)的模拟电路故障诊断方法。首先,应用Haar小波分析提取被测电路时域响应信号的小波系数作为特征参量,并生成样本数据;然后,基于样本数据,应用量子粒子群算法对GMKL-SVM的参数进行优化,并以此建立基于GMKL-SVM的故障诊断模型,用于区分模拟电路的各个故障。实例电路的单故障和双故障诊断实验结果表明,所提出的GMKL-SVM方法能较好地实现模拟电路故障诊断,与传统的GMKL-SVM方法相比,表现出了更好的性能,获得了更高的故障诊断正确率。

A novel analog circuit fault diagnosis approach is presented by using generalized multiple kernel learningsupport vector machine （GMKL-SVM） algorithm. Firstly, the wavelet coefficients of measured time responses are generated as features by using a Haar wavelet transform. Then, wavelet features are used as samples to identify parameters for GMKL-SVM method with using quantumbehaved particle swarm optimization （QPSO） algorithm. As a result, classification model based on GMKL-SVM method is constructed to diagnosis analog circuit faults. Dignostics on both single fault and double faults demonstrate that the proposed GMKL-SVM method can obtain good diagostic performance on analog circuit fault diagnosis. Additionally, compared to the traditional GMKL-SVM method, the presented approach has higher diagnostic precision.