Fault diagnosis is significant to guarantee the safety and reliability of the machinery. Local fault will make the collected vibration signals deviate from the normal signals. In order to extract fault-related features from vibration signals, many deep learning-based methods have been proposed for machinery fault diagnosis recently. However, due to the extreme conditions (e.g., high background noise, limited labeled training samples), it is a challenging task to implement features extraction from the collected signals with non-linear and non-stationary characteristics. To implement feature extraction and noise reduction of vibration signals, this paper proposes a novel network, i.e., deep morphological shrinkage convolutional auto-encoder (DMSCAE) for gearbox fault diagnosis considering the insufficient labeled training samples. Firstly, a morphological convolutional auto-encoder is proposed for noise filtering and feature extraction. Secondly, a multi-branch structure with different structural elements (SEs) is used in the morphological layer to extract impulsive components. Finally, a soft thresholding-based shrinkage is employed to filter ineffective features, where an adaptative method is developed to adjust the threshold automatically in the back-propagation procedure. The experiments on two gearbox fault diagnosis tests are conducted to verify the performance of DMSCAE. The results indicate that DMSCAE obtains a better performance for fault diagnosis than other DNNs, e.g., efficient channel attention network (ECANet), self-calibrated convolutional network (SCNet).
29 July 2024 in IEEE Transactions on Power Electronics
Most existing methods for battery state-of-health (SOH) estimation rely on centralized training mode. However, in practical applications, it is difficult for a single user to collect sufficient battery degradation data to train a model. In addition, in order to protect data privacy, users are unwilling to share data, where centralized training mode is not the best choice. To overcome these barriers, a source-free dynamic weighted federated transfer (SF-DWFT) method for battery SOH estimation is proposed, which utilizes a distributed learning paradigm to combine multiple source clients to train a global model while protecting data privacy. First, a Gaussian mixture model is used to model the high-level features extracted by the client model, thus the feature representation can be shared without needing access to the source data; then, in order to reduce the effect of distribution differences between the source and target clients on the global model, a dynamic weighted federated aggregation algorithm is proposed according to their contributions, which is measured by calculating the modified Bhattacharya distance between the source and target clients and testing error. Finally, the effectiveness of SF-DWFT is verified on several battery datasets and it achieves a statistic estimation error of 1.12% on 18 650 lithium-ion battery datasets and 4.46% on NASA battery dataset.
Valve-regulated lead–acid (VRLA) battery module is one of the important components of the uninterruptible power supply system in a large Internet data center. Battery health monitoring and anomaly diagnosis in time is significant to ensure the safe operation of a data center. In this article, a multitask joint network (MTJNet) is proposed to perform anomaly diagnosis and inconsistency identification simultaneously. First, an unsupervised learning-based encoder-decoder structure is proposed in MTJNet for battery inconsistency identification, where only health data are required for training. Second, the other branch with a classifier is further constructed to recognize the anomaly of the battery module. Third, a multitask joint training method is used to update the parameters of the two task models in MTJNet. The reconstruction error obtained by inconsistency identification task is fed into the anomaly diagnosis task. The predicted pseudo labels by the anomaly diagnosis task are feedback to inconsistency identification task to guide the encoder-decoder to provide the discriminate features. Finally, the effectiveness of MTJNet is verified on VRLA battery modules in a large data center. The experimental results illustrate that MTJNet is a good tool for anomaly diagnosis and inconsistency identification of VRLA battery modules.
13 June 2024 in IEEE Transactions on Power Electronics
Valve-regulated lead-acid (VRLA) batteries are of paramount importance in ensuring the safety and stability of large intelligent data centers (IDCs). The health management of VRLA battery suffers from the limitation of sparse charge-discharge data and nonideal samples in the high-voltage direct current system of IDC. In this article, a novel semisupervised few-shot model, i.e., densely-connected contrastive observer (DCober), is proposed for feature self-extraction of multisource signals and few-shot diagnosis of VRLA batteries. First, a view-augmented method is proposed to implement image augmentation for multichannel temporal images, which improves the feature self-extraction performance. Second, self-supervised contrastive learning is used in DCober to implement feature representation learning on massive nonlabeled data in real industrial scenarios. Finally, a similarity-based few-shot finetuning method is proposed for finetuning enhancement on the extension of limited labeled data, which improves the fault recognition performance of DCober in few-shot scenarios. The effectiveness of DCober is validated through a real industrial application of VRLA battery in Tencent T-block IDC. The testing results show that DCober outperforms other state-of-the-art methods and obtains a recognition accuracy of 95.63% in 1-shot data scenario.