Fault diagnosis of rolling bearings using singular spectrum analysis and artificial neural networks

Singular spectrum analysis (SSA) has been employed effectively for analyzing in the time-frequency domain of time series. It can collaborate with data-driven models (DDMs) such as Artificial Neural Networks (ANN) to set up a powerful tool mechanical fault diagnosis (MFD). However, take advantage SSA more MFD, quantifying optimal component threshold should be addressed. Also, exploit managed system adaptively, variation tendency its physical parameters needs caught online. Here, we present bearing method (BFDM) based on ANN and that targets these aspects. First, multi-feature is built from pure properties distilled vibration signal system. Relied SSA, measured acceleration analyzed cancel high-frequency noise. The remaining components part building establish database training ANN. Optimizing number kept then carried out obtain dataset called Tr_Da. Based Tr_Da, receive (OANN). In next period, at each checking time, another Test_Da online following same way compared result between encoded output OANN corresponding input provides bearing(s) health information. An experimental apparatus evaluate BFDM. obtained results reflect positive effects method.