Wheel bearings are essential mechanical components of trains, and fault detection of the wheel bearing is of great significant to avoid economic loss and casualty effectively. However, considering the operating conditions, detection and extraction of the fault features hidden in the heavy noise of the vibration signal have become a challenging task. Therefore, a novel method called adaptive multi-scale AVG-Hat morphology filter (MF) is proposed to solve it. The morphology AVG-Hat operator not only can suppress the interference of the strong background noise greatly, but also enhance the ability of extracting fault features. The improved envelope spectrum sparsity (IESS), as a new evaluation index, is proposed to select the optimal filtering signal processed by the multi-scale AVG-Hat MF. It can present a comprehensive evaluation about the intensity of fault impulse to the background noise. The weighted coefficients of the different scale structural elements (SEs) in the multi-scale MF are adaptively determined by the particle swarm optimization (PSO) algorithm. The effectiveness of the method is validated by analyzing the real wheel bearing fault vibration signal (e.g. outer race fault, inner race fault and rolling element fault). The results show that the proposed method could improve the performance in the extraction of fault features effectively compared with the multi-scale combined morphological filter (CMF) and multi-scale morphology gradient filter (MGF) methods.
Intelligent mechanical fault diagnosis algorithms based on deep learning have achieved considerable success in recent years. However, degradation of the diagnostic accuracy and operational speed has been significant due to unfavorable working conditions and increasing network depth. An improved version of ResNets is proposed in this paper to address these issues. The advantages of the proposed network are presented as follows. Firstly, a multi-scale feature fusion block was designed, to extract multi-scale fault feature information. Secondly, an improved residual block based on depthwise separable convolution was used to improve the operational speed and alleviate the computational burden of the network. The effectiveness of the proposed network was validated by discriminating between diverse health states in a gearbox under normal and noisy conditions. The experimental results show that the proposed network model has a higher classification accuracy than the classical convolutional neural networks, LeNet-5, AlexNet and ResNets and a faster calculation speed than the classical deep neural networks. Furthermore, a visual study of the different stages of the network model was conducted, to effectively comprehend the operational processes of the proposed model.
Fault detection and diagnosis techniques for rotating mechanical components are crucial for the safety, efficiency, reliability of mechanical systems. In recent decades, many sparse representation approaches based on parametric dictionary design have been successfully applied to extract fault features from vibration signals. However, most of them rely on the classical correlation filtering algorithm (CFA), which has some shortcomings, such as poor antinoise ability and an extensive computation load. To address these issues, this paper proposes a novel water cycle algorithm (WCA)-optimized fault impulse matching algorithm (FIMA) for parametric dictionary design, which can match the underlying fault impulse structure of fault signals by applying a comprehensive strategy of local matching and global matching. With the proposed method, the accuracy of wavelet parameter identification and calculation efficiency are improved significantly. In addition, the method is suitable for constructing parametric dictionaries with different wavelet bases according to different rotating machinery components. The effectiveness of the proposed method is verified by the simulated signals, as well as the practical wheelset bearing faults (outer race fault and inner race fault) and gearbox (broken tooth) signals. The results of a comparison study show that the proposed method outperforms the classical CFA, K-SVD and whale optimization algorithm-optimized orthogonal matching pursuit in weak fault feature extraction.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.