In motor fault diagnosis technique, the detections of vibration and stator current frequency components are two main detecting means. This article discusses the detection method of the vibration fault signal. Because this signal is a non-stationary random signal, the fault signals often contain a lot of time-varying, burst properties, the traditional Fourier signal analysis can not effectively extract the motor fault characteristics, it is likely that the weak signal of the rich failure information is regarded as noise to be deleted. For this the wavelet packet transform is used to extract the fault characteristics of the signal information. The result obtained is taken as the neural network input signal, L-M neural network optimization method is used for training, and then, the BP network is used for fault recognition. It also uses the Matlab software to carry out the simulation. It confirms that the method is valid for the motor fault diagnosis and the diagnosis is accurate.%在电动机故障诊断技术中,基于振动和定子电流频率成分的检测是电动机故障检测的两种主要手段.讨论了基于振动故障信号的检测方法.由于电动机振动信号是非平稳随机信号,故障信号中往往含有大量的时变、短时突发性质的成分,传统的傅里叶信号分析不能有效地提取电动机的故障特征,而且还可能将含有丰富故障信息的微弱信号作为噪声去除.因此,引入比小波分析更强的小波包变换技术来提取信号的故障特征信息,得到的结果作为神经网络的输入信号,用神经网络的L-M优化算法来进行训练,然后用BP神经网络来进行故障识别.采用Matlab软件进行仿真,证实该方法对电动机故障诊断的有效性和准确性.
展开▼