Intelligent fault diagnosis (IFD) is important for rotating machinery maintenance. Unfortunately, fault diagnosis training models often degenerate if unknown domain shifts exist between different working conditions when performing IFD. To deal with this problem, more generalized features related to rolling element bearing faults should be explored so that the generalized capacity of the training model is boosted for unseen target domain data. In this paper, a new algorithm using pseudo domain labels is proposed to explore subdomain distributions within each subdomain at the domain level. The idea behind the proposed method is that the domain shifts caused by variable working conditions, like varying speeds, should also be considered since the data may show a dynamic distribution of temporal features that are not limited to spatial distributions. That is, the original domain distribution could be further divided into several latent subdomains by introducing pseudo domain labels, which enables the proposed method to learn domain specific features. Furthermore, the diversity of learned features across subdomains ensures comprehensive feature coverage during model training, while the inherent similarities between these domains enhance the capacity of the model for domain generalization. To figure out how the domain label updates, a domain-class label is initially introduced to facilitate fine-grained feature learning, enabling the model to capture as many features as possible. Then an adversarial learning strategy is employed to separate the domain and class information. Specifically, pseudo domain labels are determined using class invariant features, while class labels are distinguished using features that are invariant across multiple latent subdomains. These two steps are equivalent to a min–max game, like adversarial learning. By exploring features from the class and domain levels, the domain generalization capabilities of the model can be improved, thereby further increasing the accuracy of results. Experiments on two public bearing datasets show that the proposed method outperforms state-of-the-art methods. Additionally, by limiting the number of accessible data from known source domains, the proposed method shows the potential to maintain satisfactory domain generalization capacities when combined with few-shot learning.

Unknown domain shift caused by the unavailability of target domain during training phase degrades the performance of intelligent fault diagnosis models in practical applications. Domain generalization (DG)-based methods have recently emerged to alleviate the influence of domain shift and improve the generalization ability of models toward invisible working conditions. However, most existing studies are conducted on multiple fully labeled source domains. Meanwhile, domain-specific information related to the variations of working conditions is often neglected during model training. Therefore, in order to realize reliable generalization fault diagnosis based on partially labeled source domains, this article proposes a contrast-assisted domain-specificity-removal network (CDSRN) to extract transferable features from domain-specificity-removal perspective. Concretely, a domain-specific feature removal branch is designed to disentangle domain-invariant features and domain-specific features, thus excavating generalized information only in domain-invariance dimension. Simultaneously, proxy-contrastive representation enhancement module is embedded to facilitate the fault class-discriminative and domain-discriminative feature learning, thereby assisting the model in further improvement of generalization capability. Experimental studies confirm the effectiveness and competitiveness of the proposed CDSRN in semi-supervised generalization fault diagnosis.

Multi-source domain adaptation (MSDA) has demonstrated superior performance in intelligent fault diagnosis (IFD) compared to single-source domain adaptation (SSDA), as it can provide more comprehensive and diverse information from multiple fully-labeled source domains. However, in many real industrial scenarios, acquiring multiple fully-labeled source domains is challenging because labeling all the source domains is as expensive and laborious as labeling the target domain. Given this concern, a cross-supervised multisource prototypical network (CSMPN) is proposed for multi-source few-shot fault diagnosis. Specifically, a domain-shared and a domain-individual branch are constructed to realize shared domain alignment across all the source and target domains and individual domain alignment of source-target domain pairs, respectively. Within two branches, domain alignment is realized by the designed prototypical contrastive learning (PCL) module. In the PCL module, we propose a prototype calibration strategy to address the issue of biased prototype estimation owing to outlier samples. In addition, a two-stage pseudo-labeled sample selection mechanism is proposed to enhance the feature representation ability of two branches. At the end of the two branches, we design a cross-supervised learning (CSL) module to realize mutual and collaborative learning between the two branches, which can further improve the diagnosis performance on the target domain. Experiments on two different bearing datasets are implemented to verify the superiority of the proposed method compared with the comparison methods. Our code is available at https://github.com/YNWA-Zhang/CSMPN .

In recent years, the surge of artificial intelligence has propelled autonomous driving technology to the forefront, capturing growing interest and enthusiasm. As a sub-module within autonomous driving, research in traffic sign detection and recognition has significantly advanced with a growing focus on utilizing deep learning methods. Nevertheless, in complex real-world road scenarios, traffic signs often exhibit a long-tailed distribution, with the majority of instances concentrated in a few frequent categories and a scarcity in the remaining ones. Considering conventional Traffic Sign Detection and Recognition methods are crafted using manually curated datasets, the class imbalance may detrimentally impact the efficacy of the detection model. In this paper, we first propose a gradient-guided loss reweighting model that dynamically reweights the loss for positive and negative samples based on the cumulative gradients across each category. Additionally, a classification bias-based refinement module is proposed to fine-tune these weights during training, based on the false positive and false negative rate. This serves to suppress a drop in precision for tail categories resulting from the gradient-guided loss reweighting module, thus further balancing the entire training process for improved results. Extensive experiments are performed on the TT100K and GTSDB datasets, yielding significant advancements surpassing state-of-the-art methods.

Abstract The problem of finding the maximum independent sets (or maximum cliques) of a given graph is fundamental in graph theory and is also one of the most important in terms of the application of graph theory. Let $A(n,d,w)$ be the size of the maximum independent set of $Q_{n}^{(d-1,w)}$, which is the induced subgraph of points of weight $w$ of the $d-1^{th}$-power of $n$-dimensional hypercubes. In order to further understand and study the dependent set of $Q_{n}^{(d-1,w)}$, we explore its clique number and the structure of the maximum clique. This paper obtains the clique number and the structure of the maximum clique of $Q_{n}^{(d-1,w)}$ for $5\leq d\leq 6$. Moreover, the characterizations for $A(n,d,w)=2$ and $3$ are also given.

Aluminum electrolytic capacitors (AECs) are utilized as the key components in power electronic transformers (PETs). The AEC degradation monitoring is crucial for the maintenance of PET. Degradation of AEC performance is often reflected by changes of capacitance (C) and equivalent series resistance (ESR). SinceCand ESR dominate the low- and mid-frequency impedance characteristics of the AEC, respectively, the features of the corresponding frequency bands of the respective signals need to be simultaneously extracted. However, the current studies on parameter identification of AECs have not focused on this issue. In this article, the double-scale convolutional autoencoder and extreme learning machine (DCAE-ELM) framework is proposed to identifyCand ESR based on AEC voltage. Specifically, DCAE extracts the low- and mid-frequency features of AEC voltages with large- and small-scale convolutional kernels, respectively. Then, ELM is employed to identifyCand ESR based on the features extracted by DCAE. Moreover, the mathematical mechanisms between the gradients and reconstructed data of DCAE with data concatenated in columns (cDCAE) and rows (rDCAE) are analyzed. Validation results of both simulation and experimental data have verified the data reconfiguration performance of rDCAE and the parameter identification capability of the proposed DCAE-ELM framework.

Traffic sign detection and recognition (TSDR) has attracted extensive studies recently due to its broad application prospect in Intelligent Transport Systems. TSDR is still challenging due to the small size of traffic signs in the image. Besides, the traffic signs in the real world exhibit a long-tailed distribution (i.e., data for most categories are scarce while for others are abundant.), which will lead to a significant performance drop of the detection framework. In this paper, we propose a novel traffic sign detection framework to address these challenging problems. In order to detect small traffic signs, we propose an effective adaptive and attentive spatial feature fusion module which learns the spatial attention map to fuse different feature maps at each scale while emphasizing or suppressing the features at different regions. This module can significantly alleviate the inconsistency among features and enhance feature representations of small objects. Furthermore, to address the long-tailed data problem, a hierarchical group softmax head which constructs a label tree to divide categories into different groups is proposed, in this way, categories in each group have relatively similar frequencies, then the softmax is applied in each relatively balanced group to calculate the probability of each category. Extensive experiments conducted on the TT100K and GTSDB datasets demonstrate that the proposed method achieves notable improvement in both the small traffic signs and long-tailed detection problems in TSDR.