A Novel Rolling Bearing Fault Diagnosis Method Based on BLS and CNN with Attention Mechanism

In actual industrial application scenarios, noise pollution makes it difficult to extract fault features accurately via conventional methods. A novel method for rolling bearing diagnosis combining a convolutional neural network (CNN), an attention mechanism squeeze-and-excitation (SENet) module and broad learning system (BLS) is proposed (SECNN–BLS). The one-dimensional vibration signal processed by using multiple short-time Fourier transforms (STFT); the two-dimensional image in time-frequency domain used as model input. CNN feature extraction process, SENet introduced replace convolution layer, global information obtained through squeeze operation. Excitation operation captures importance of channels, assigns weights adaptively improve on important eliminates interference irrelevant without increasing spatial temporal complexity. weighted representation then transferred BLS has characteristics flat structure ridge regression quickly solve weights; classifier, can save more computing resources accuracy classification. SECNN–BLS achieved than 98% Society Machinery Failure Prevention Technology (MFPT) dataset. We also demonstrate excellent performance noisy environment.