基于软阈值与自编码器的滚动轴承异常检测

Anomaly detection of rolling bearings based on soft threshold and autoencoder

滚动轴承作为一种通用的机械部件在旋转机械设备中被广泛使用,由于滚动轴承经常受到交变载荷作用,容易发生故障。振动信号是机械设备监测和诊断常用的信号数据,滚动轴承在不同工况和健康状况下其振动信号在时域以及频域都会发生相应的变化。本文提出了自编码器与软阈值结合利用快速傅里叶变换(FFT)得到的频域信息对轴承进行异常检测方法。实验结果表明,在测试样本中通过该方法提取数据特征能够较为有效地检测出轴承的异常状态,相较于一分类支持向量机、排列熵等典型异常检测方法,在辛辛那提IMS轴承全寿命数据集中,该方法诊断效果稳定,对于早期故障有更高的灵敏性。

In the industry,rolling bearings are widely used in rotating machinery as a general mechanical component.Rolling bearings are often subject to alternating loads and thus are prone to failure.Vibration signals are commonly used signal data for mechanical equipment monitoring and diagnosis.Under different working conditions and health conditions,the vibration signals of rolling bearings can change accordingly in the time domain and frequency domain.Therefore,an anomaly detection method for bearings using the frequency domain information obtained by the fast Fourier transform(FFT)combined with the autoencoder and the soft threshold is proposed in this paper.The experimental results show that extracting data features from the test samples by this method can effectively detect the abnormal state of the bearings.Compared with typical anomaly detection methods such as the one-class support vector machine,and permutation entropy,etc.,the method used in Cincinnati IMS bearing life-cycle data set has stable diagnosis effect,and is more sensitive to early faults.