A class alignment method based on graph convolution neural network for bearing fault diagnosis in presence of missing data and changing working conditions

Health monitoring in rotatory machinery is a process of developing a mechanism to determine its state of deterioration. It involves analysing the presence of damage, locating the fault, determining the severity of the problem, and calculating the amount of time that the machine can still be used effectively by making use of signal processing methods. The journey started to repair when the machine fails and progressed to the modern era, which involves the use of advanced sensors to capture data and conduct on-line signal processing methods to extract relevant features. By seamlessly integrating advanced smart sensing, data collection, and intelligent algorithms, modern technologies have transformed the landscape of condition-based maintenance for rotary machinery, bridging the gap between fundamental understanding and practical engineering applications. In this review paper, first, the roadmap of the condition-based maintenance (CBM) journey for rotary machinery is briefly introduced. Then, CBM task techniques are reviewed in the context of manual identification of defects, applying artificial intelligence (AI) model to identify the defect in the rotary machinery, and AI to carry out defect prognosis and determine the remaining useful life. Finally, the challenges, and issues of signal processing methods to detect faults in rotary machinery, and remedies to overcome such challenges are deeply discussed and future research directions are identified to ensure safe operation for rotary machinery.

Section: Graph Convolutional Network (Gcns)mentioning

confidence: 99%

Latest innovations in the field of condition-based maintenance of rotatory machinery: a review

Kumar,

Gandhi,

Tang

et al. 2023

“…The local maximum mean discrepancy (LMMD) was introduced into the domain adaptation network by Ghorvei et al [70], to reduce the discrepancies of feature structure between subdomain and global domain. On this basis, Kavianpour et al [71] improved the LMMD by adopting the MK-LMMD, which aligns the features for intra-class discrepancies. In addition, considering that a single distribution cannot well represent the discrepancies of the feature distributions, Xiao et al [53] introduced a conditional distribution, and constructed a loss function based on joint maximum mean discrepancy to achieve feature alignment.…”

Section: Introductionmentioning

confidence: 99%

DDAGCN: an unsupervised cross-domain identification method for tie rod bolt loosening in a rod-fastening rotor system under different working conditions

Zhou,

He,

Yang

et al. 2024

The cross-domain diagnosis of tie rod bolt loosening is essential for guaranteeing the healthy operation of rod-fastening rotor (RFR) systems. The unsupervised domain adaptation (UDA) method effectively alleviates the impact of domain discrepancy and has been applied for cross-domain diagnosis. Traditional UDA methods mainly focus on the marginal and conditional distributions with fixed weights to adapt the domain distribution discrepancy. However, the fixed distribution combination cannot satisfy the requirement of feature domain alignment under different working conditions, and the relative importance of the two distributions cannot be evaluated quantitatively. This paper proposes an improved dynamic distribution adaptive graph convolutional network (DDAGCN) for the cross-domain diagnosis of tie rod bolt loosening under variable working conditions. This method can quantitatively evaluate the relative significance of each distribution in representing the distribution discrepancy. First, it combines the convolutional neural network (CNN) and the graph convolutional network (GCN) to extract the features in the graph structure by using the connection relationship between nodes, and realizes the full extraction of local neighbourhood information of nodes. Then, the dynamic distribution adaptive alignment strategy is introduced to construct the dynamic linear combination of marginal and conditional distributions, so as to measure the distribution discrepancy between domains. Meanwhile, the domain adversarial module is combined to further reduce the domain gap and finally realize feature alignment. The extracted domain invariant features can effectively enhance the generalization ability and fault identification ability of the model. The case of the public bearing dataset verifies that the effectiveness and generalization ability of the proposed method for cross-domain fault diagnosis under variable working conditions is superior to other compared methods. In addition, the identification ability of the proposed method for the degree of tie rod bolt loosening is verified by the self-made bolt loosening dataset of the RFR system.

“…As a core component of rotating machinery equipment, rolling bearing is easily damaged by improper maintenance, frequent working condition switching, mechanical fatigue, and other factors, and then causes equipment failure [1,2]. According to statistics, nearly half of the failures in rotating machinery are caused by rolling bearings [3].…”

Section: Introductionmentioning

confidence: 99%

Novel imbalanced subdomain adaption multiscale convolutional network for cross-domain unsupervised fault diagnosis of rolling bearings

Huo,

Deng,

Zhang

et al. 2023

Data on the vibration signals collected from rolling bearings mostly belongs to health conditions, leading to an imbalanced data distribution. In addition, frequent switching of operating conditions results in unlabeled data collected under a specific working condition. To address these challenges, this paper proposes a novel network for cross-domain unsupervised fault diagnosis of rolling bearings considering the imbalanced data. First, a multiscale parallel features extraction (MPFE) is developed, which can fully mine each fault type's inherent invariant domain features with a high value for fault identification from original data. Second, a squeeze-and-excitation attention mechanism (SEAM) is constructed to enhance features conducive to model classification and suppress redundant features. Finally, a new subdomain adaption algorithm is presented to adaptively align the relevant imbalanced subdomain data of two domains. The proposed method was validated on multiple unsupervised cross-domain diagnostic tasks on two bearing datasets. Experimental results manifest that the proposed method has stable generalization performance and excellent robustness.