A hybrid intelligent rolling bearing fault diagnosis method combining WKN-BiLSTM and attention mechanism

Abstract

Fault diagnosis of rolling bearings helps ensure mechanical systems’ safety. The characteristics of temporal and interleaved noise in the bearing fault diagnosis data collected in the industrial field are addressed. This paper proposes a hybrid intelligent fault diagnosis method (WKN-BiLSTM-AM) based on WaveletKernelNetwork (WKN) and bidirectional long-short term memory (BiLSTM) network with attention mechanism (AM). The WKN model is introduced to extract the relevant impact components of defects in the vibration signals, reduce the model training parameters and facilitate the processing of signals containing noise. Then, the fusion of spatial-temporal features is achieved by combining BiLSTM networks to compensate for the lack of individual networks that ignore the dependent information between discontinuous sequences. Finally, the AM module is introduced to improve the feature coding performance of BiLSTM and fault diagnosis accuracy. Comparison and validation between the proposed WKN-BiLSTM-AM method and other state-of-the-art models are given on the Case Western Reserve University (CWRU) and Paderborn University (PU) datasets. The experimental results verify the effectiveness of the proposed model in bearing fault diagnosis, and the model’s generalization capability.

A deep learning based health indicator construction and fault prognosis with uncertainty quantification for rolling bearings

“…In summary, the calculation of mean output and model variance are defined in equations ( 9) and (10), respectively. Meanwhile, the prediction error term σ 2 e is acquired through equations ( 15) and (16).…”

“…In recent years, the fault prognosis of rolling bearings has received an increasing amount of attention. Its health status straightly affects the reliability and safety of the rotary machinery [10]. Moreover, the fault prognosis of rolling bearings is of vital importance because an accurate fault prognosis of rolling bearings can detect the degradation trend previously and reform the maintenance strategies.…”

SO-IMCKD processed signal improving MSCNN model’s fault diagnosis accuracy for drilling pump fluid end

“…And CNN mainly consists of the input layer, convolutional layer, activation function, pooling layer, fully connected layer, and output layer [30][31][32]. In CNN, the kth convolutional feature p l k [ẋ, ẏ] of the lth layer can be expressed as shown in equations ( 19) and (20), respectively…”

“…Song and Jiang [18] transformed the time series signals in the chemical process into matrix graphs and input them into multi-scale convolutional neural networks (MCNNs), realizing various fault diagnosis in the chemical production process, with a recognition accuracy of 88.54%. Some scholars have also improved CNN to improve the diagnostic effect of bearings [19,20]. Research has also found that CNN can effectively extract images generated by continuous wavelet transform (CWT) [21][22][23][24], and realize the diagnosis of different conditions in different fields.…”

A hierarchical transformer-based adaptive metric and joint-learning network for few-shot rolling bearing fault diagnosis

“…In addition to the design of the classifier, the structure of the encoder also affects the information extraction ability of the model. Currently, the mainstream feature extraction networks for fault diagnosis mostly consist of variants of convolutional neural networks (CNN) [25,26] or recurrent neural networks (RNN) [27,28]. These architectures adopt a sequential processing approach for modeling, often requiring the stacking of multiple layers or the accumulation of information over several time steps to capture long-range interdependencies.…”