应用模糊最优小波包和LS-SVM的模拟电路诊断 .pdfVIP

学兔兔 第31卷 第6期 仪 器 仪 表 学 报 Vol_31 No．6 2010年6月 Chinese Journal of Scientific Instrument Jun．2010 应用模糊最优小波包和LS—SVM的模拟电路诊断 王佩丽，彭敏放，杨易曼，吴俊丽，何怡刚 (湖南大学电气与信息工程学院 长沙 410082) 摘 要：为解决模拟电路故障诊断中的特征提取困难并对模拟电路故障信号进行有效的分类，提出了一种结合模糊理论、小波 包分解和最小二乘支持向量机(LS—SVM)的模拟电路诊断方法。该法首先对模拟电路的响应信号进行小波包分解，并引入模糊 准则对其优化，得到由分类能力强的最优小波包基能量值构成的特征集，然后将特征集输入LS—SVM网络，实现对不同故障类 型的识别。小波包的优化分解减小了LS—SVM网络的规模，从而降低了算法复杂度，加快了网络的训练时间和分类速度。模拟 诊断实例表明，此方法能快速有效地实施模拟电路的故障定位。 关键词：故障诊断；模拟电路；最优小波包分解；模糊规则；最小二乘支持向量机 中图分类号：TM131 文献标识码：A 国家标准学科分类代码：470．40 Analog circuit diagnosis using fuzzy-rule based optimal wavelet packet and LS-SVM Wang Peili，Peng Minfang，Yang Yimin，Wu Junli，He Yigang (College of Electrical and Information Engineering，Hunan University，Changsha 412008，China) Abstract：In order to solve the difficulty of feature extraction and classify fault signals effectively in analog circuit fault diagnosis，a new analog circuit diagnosis method based on optimal wavelet packet combined with fuzzy—rule and least square support vector machine(LS—SVM)is proposed in this paper．The response signals of the analog circuit are preprocessed using wavelet packet transform and the optimal wavelet packet coefficients with better classification capacity are obtained using the fuzzy rules．Then，the feature set that is composed of optimal wavelet packet energy is inputted into a LS-SVM network to identi~different fault cases．The algorithm of optimal wavelet packet transform combined with fuzzy—rule decreases the size of LS—SVM network and the complexity of the algorithm while improves