基于词袋模型和极限学习机的轴承故障诊断

Bearing Fault Diagnosis Based on BoW and ELM

特征提取为设备故障诊断的关键步骤,现有特征提取主要基于时域、频域和时频域方法,但在噪声大、混合故障多发、变转速等复杂工况下应用受限。提出一种基于词袋模型(BoW,bag of words)的振动信号故障诊断方法。该方法改变了传统基于一维信号进行故障诊断的模式,首先将一维振动信号转换成二维灰度图像信号,结合图像处理的方法来实现故障诊断。同时针对一维振动信号升维的过程中忽略了信号的空间特性这一不足,引入极限学习机对算法优化,并以滚动轴承为研究对象验证了方法的有效性。

Feature extraction is a key step in equipment fault diagnosis.The existing methods are mainly based on time domain,frequency domain and time-frequency domain,but their usage is limited in the actual conditions such as high noise,mixed fault and unfixed rotation speed.A method for bearing fault diagnosis based on bag of words（BoW）,which differs from traditional one-dimensional signal based fault diagnosis method,is put forward.The vibration signal will first be transformed into two-dimensional gray-scale image signal and then will be processed by using image processing method.Extreme learning machine（ELM） will be used to optimize the algorithm on account of the negligence of the signal＇s spatial characteristic.The effectiveness of the method is verified by using the rolling bearing as the research object.