基于小波变换和神经网络的模拟电子电路故障诊断  

Fault diagnosis of analog electronic circuits based on wavelet transform and neural network

作　　者：郑磊 蒋玮[1] 胡仁杰[2] 张震宇 黄慧春[2] Zheng Lei;Jiang Wei;Hu Renjie;Zhang Zhenyu;Huang Huichun(School of Electrical Engineering,Southeast University,Nanjing 210096,China;National Demonstration Center for Electrical and Electronic Experimental Teaching,Southeast University,Nanjing 210096,China)

机构地区：[1]东南大学电气工程学院,江苏南京210096 [2]东南大学电工电子国家级实验教学示范中心,江苏南京210096

出　　处：《南京理工大学学报》2024年第3期310-317,共8页Journal of Nanjing University of Science and Technology

摘　　要：电路故障是影响模拟电子电路实验进度和实验效果的重要因素,为实现对故障快速、准确的分类和诊断,提出了一种基于小波变换和神经网络的故障诊断方法。首先,分析了模拟电子电路的实验进程和电路特点,以电路激励和响应信号为切入点,提出了融合小波分解和交叉小波的特征提取方法,获取了包含信号频域特征和相位差信息的特征向量。然后,设计基于L-M优化的神经网络算法,将降维和正则化处理后的特征向量作为模型的输入参数进行训练,得到最终的诊断结果。最后,在共发射极放大电路故障集上的测试结果表明,所提方法能够有效提高不同电路故障诊断的速度和准确率。Circuit fault is an important factor affecting the experimental progress and effect of analog electronic circuits,and in order to realize the rapid and accurate classification and diagnosis of faults,a fault diagnosis method based on wavelet transform and neural network is proposed.Firstly,the experimental process and circuit characteristics of analog electronic circuits are analyzed,and a feature extraction method combining wavelet decomposition and cross wavelet is proposed taking the circuit excitation and response signals as the starting point,and the feature vectors containing the signal frequency domain features and phase difference information are obtained.Then,a neural network algorithm based on L-M optimization is designed,and the feature vectors after dimensionality reduction and regularization are trained as the input parameters of the model,and the final diagnostic results are obtained.Finally,the test results on the fault set of the common emitter amplification circuits show that the proposed method can effectively improve the speed and accuracy of fault diagnosis of different circuits.

关 键 词：小波变换 交叉小波 神经网络 故障诊断 

分 类 号：TN710[电子电信—电路与系统] TP183[自动化与计算机技术—控制理论与控制工程]