摘要
提出了采用小波变换和独立成分分析(ICA)作为预处理器来进行特征提取的神经网络开关电流电路故障诊断方法。该方法对采集到的故障响应信号进行Haar小波正交滤波器分解,获得低频近似信息和高频细节信息;然后利用独立成分分析方法进行ICA故障特征提取;最后将所得到的最优故障特征输入到BP神经网络中进行故障分类。对六阶切比雪夫低通滤波器和六阶椭圆带通滤波器电路进行了仿真实验验证,获得了100%的故障诊断准确率,与其他方法进行比较,实验结果显示了该方法的优越性。
A neural-network switched-current circuit fault diagnosis approach utilizing wavelet transform and ICA feature extraction as preprocessors for feature extraction is proposed. The diagnostic approach performs Haar wavelet transform（ HWT） on the acquired fault response signals,and the low frequency approximation information and high frequency detail information are obtained,then ICA fault feature is extracted by employing independent component analysis method. Finally,the obtained optimal fault features are sent to BP neural network to classify different faults. A 6 th-order Chebyshev low-pass filter circuit and a6 th-order Elliptic bass-pass filter circuit were used to conduct simulation experiment and verify the proposed method,and the fault diagnosis accuracy of 1 0 0 % is achieved. Compared with other methods,the experiment result indicates the superiority of the presented method.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2015年第10期2389-2400,共12页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(61201108
61501162)
中国博士后科学基金(2014M551797
2015T80650
2015M571926)项目资助
