Streaming data of machines is continuously collected in practical applications, which produces new fault information with respect to the health change. Therefore, a lifelong-learning intelligent diagnosis model is desired for new fault type recognition based on the streaming data. However, existing research in intelligent fault diagnosis always treats new fault type detection and class incremental learning as two independent problems, which reduces their practicality in industrial applications. To tackle this limitation, a graph continual learning network is constructed for incremental intelligent diagnosis of new faults. The method integrates the advantages of both new fault type detection and class incremental learning. In the method, a graph convolutional network (GCN) based model is formulated for detecting new classes to prejudge whether the DL model needs to be updated. Once any new class is detected, class incremental learning is started automatically to update the DL model without leading to catastrophic forgetting. The proposed method is applied to a pump fault diagnosis case with incremental fault types. Results show that the proposed method offers an effective solution for online intelligent fault diagnosis with satisfactory classification performanceNote to Practitioners—Existing DL-based intelligent diagnosis models often assume the closed-set assumption, i.e., fault types of the monitoring data are the same as those of the training data. In the situations where the assumption is held, DL models receive high-precision recognition results towards instances from the monitoring data stream. However, when a new fault pattern appears in the monitoring data stream, the previous well-trained diagnosis model will inevitably classify the instances into one of the known patterns, resulting in untrustworthy results. This paper proposes a method with the aim of tackling the above issue. The method is able to realize the following functions: Once the instances in the streaming data are detected as a new fault pattern, the class incremental learning will be started automatically. The detected class is used to update the diagnosis model. On the contrary, if the instances are not detected as a new class, they will be sent to the diagnosis model for fault recognition. An incremental diagnosis task is designed under a centrifugal pump application scenario. The results demonstrate the feasibility of the proposed method. The method is anticipated with more application scenarios to verify its superiority.

Intelligent fault diagnosis is crucial for ensuring equipment reliability and preventing costly failures in industry production. Fault diagnosis relies on contact-based measurement techniques in most cases, which exhibit inherent limitations under some extreme conditions. Meanwhile, practical industrial applications increasingly demand systems with low power consumption and low computational complexity, alongside intelligent algorithms for cross-domain generalization. In this study, to overcome limitations associated with contact-based fault diagnosis methods, event-based cameras with dynamic vision are proposed to measure subtle vibrations and diagnose faults for rotating machinery. A hybrid method that integrates meta-learning for domain generalization and deep correlation alignment is proposed to enhance cross-domain diagnosis accuracy under previously unseen operational conditions. Representation and preprocessing methods are employed to augment the dynamic vision data and transform the sparse, binary event-based signals into spatiotemporal frames. The conversion enables the dynamic vision data to be processed by the second-order leaky integrated-and-fire model, supported by neuromorphic computing. That not only improves feature extraction performance but also leverages the inherent low power consumption and computational efficiency of neuromorphic computing systems. Experiments are conducted to evaluate the proposed method, confirming the promising performance in unknown working conditions and its potential for industrial applications.

The construction of a digital twin model for the full life cycle of rolling bearings is of great significance for analyzing their degradation performance and health management. However, existing researches primarily concentrate on the degradation of the outer ring of bearings. The cage, as an important component of bearings, lacks extensive research. Therefore, this article proposes a digital twin assisted assessment method for the degradation of bearing cages. First, a dynamic model including bearing cage fracture is established to generate simulation degradation signals. Second, the simulation signal is modified based on the squeeze and excitation cycle generative adversarial network (SECycleGAN) to minimize the characteristic distribution differences between the simulation and real signals. Finally, the corrected high-fidelity signal is used to train the proposed selective kernel transformer (SKformer) model to assess the degradation stage of the bearing cage. This model can simultaneously capture the long-range temporal correlation features and local mutation multiscale features of the input signals, thus improving the model's recognition ability and generalization performance. The effectiveness of the proposed method is demonstrated through signals collected on real and open-source bearing cage degradation test rigs. The results indicate that the proposed method can produce high-fidelity bearing cage degradation signals and achieve better classification accuracy with limited data.

Accurate machinery health prognosis, also known as remaining useful life (RUL) prediction, is critical for preventing catastrophic accidents and implementing predictive maintenance strategies. This makes it a highly attractive research area. Many existing studies have been developed on unimodal data, yet such data can only provide a restricted perspective and incomplete health state monitoring. Some researchers seek to address this issue from a multimodal standpoint. While promising, these methods still have certain shortcomings: 1) the imbalance for unlabeled and low-quality data compared to well-labeled data is not considered, causing their potential underexploited; 2) information richness during fusion is insufficient, discarding many valuable original and subtle health state cues, and they fail to timely tackle unexpected online anomalies; and 3) correlations and complementary information across modalities are neglected. To address these challenges, a multimodal correlation-aware fusion framework is proposed for machinery health prognosis. The framework adopts a pretrain-finetune paradigm with two parts. The first part achieves effective exploitation of the unlabeled and low-quality multimodal data pieces. The second part, through degradation pattern recognition, enables the framework to bridge the gap between scarce multimodal labeled data and accurate RUL prediction. A real industrial multimodal dataset of milling cutters is applied to demonstrate the proposed framework. Results from a series of ablation experiments and comparisons with state-of-the-art prediction methods indicate the effectiveness of each key component within the framework and its overall superiority. The framework shows promise in adapting to more downstream industrial tasks, providing accurate and reliable insights from limited data resources.

With the continuous development and advancement of neural networks, in the application of neural networks, users not only require neural networks to be able to complete a given task but also want to know when they can trust the network’s prediction results and when they need to be cautious about the prediction results. In response to the need for uncertainty estimation of neural networks, many researchers have invested in the study of uncertainty estimation. Existing uncertainty evaluation methods are difficult to apply to deep neural networks with large parameter scales, complex internal structures, and mappings between inputs and outputs that are hard to express. This paper proposes a key nodes based resampling uncertainty estimation method ((KN-RUE), which achieves uncertainty estimation of prediction results for arbitrarily given large-scale neural networks. In this method, the first step involves analyzing the differences in feature space between adversarial and clean samples, identifying the main nodes affected by adversarial samples, and determining the critical nodes within the network. Next, by resampling the parameters of key nodes, the model is extended while ensuring model performance as much as possible, thus completing the measurement of uncertainty in prediction results. Through experiments, the effectiveness of the extended model and the superiority of uncertainty estimation performance in KN-RUE have been verified.

Event-based cameras are the emerging bioinspired technology in vision sensing. Different from the traditional standard cameras, the event-based cameras asynchronously record the brightness change per pixel, and have the great merits of high temporal resolution, low energy consumption, high dynamic range, etc. While the event-based cameras have been initially exploited in several common vision-based tasks in the recent years, the investigation on machine condition monitoring problem is quite limited. This article offers the first attempt in the current literature on exploring the contactless event vision data for machine fault diagnosis. A vibration event representation is proposed to transform the event records into typical data samples, and a deep convolutional neural network model is used for processing the event information. To enhance the model robustness against environmental noisy vision events, an event data augmentation method is proposed to introduce variations of the event patterns. A deep representation clustering method is further proposed to improve the pattern recognition performance with respect to different machine health conditions. Experiments on the event vision-based rotating machine fault diagnosis problem are carried out. It is extensively validated that high fault diagnosis accuracies can be obtained using the vision data from the event-based cameras, which are competitive with the popular accelerometer data. Considering the properties of flexibility, portability, and data recognizability, the event-based cameras thus provide a promising new tool for contactless machine health condition monitoring and fault diagnosis.

Machinery often operates under time-varying conditions, which can lead to distribution discrepancies in degradation samples. However, most existing domain generalization-based methods for predicting remaining useful life (RUL) are applied to constant operation conditions, and they may demonstrate performance deteriorations across alternating operation conditions. This article proposes an optimal-subdomain generalization method for RUL prediction. It considers the local fluctuations of monitoring signals under time-varying operation conditions and performs subdomain generalization to overcome the distribution discrepancies of samples. First, the RUL labels are discretized into pseudolabels representing different health states. To guide the discretization, an optimal discretization strategy is proposed to establish the relationship between the generalization error and degradation samples. Second, the optimal subdomain generalization model is utilized to extract subdomain representations among different operation conditions for the estimation of RULs. Finally, the run-to-failure experiments on hub-bearings are conducted for demonstration. The prediction results with high accuracy show advantages of the proposed method for RUL prediction.

Due to the benefits of reduced maintenance cost and increased operational safety, effective prognostic methods have always been highly demanded in real industries. In the recent years, intelligent data-driven remaining useful life (RUL) prediction approaches have been successfully developed and achieved promising performance. However, the existing methods mostly set hard RUL labels on the training data and pay less attention to the degradation pattern variations of different entities. This article proposes a deep learning-based RUL prediction method. The cycle-consistent learning scheme is proposed to achieve a new representation space, where the data of different entities in similar degradation levels can be well aligned. A first predicting time determination approach is further proposed, which facilitates the following degradation percentage estimation and RUL prediction tasks. The experimental results on a popular degradation data set suggest that the proposed method offers a novel perspective on data-driven prognostic studies and a promising tool for RUL estimations.

Website fingerprinting attacks can infer which website a user visits over encrypted network traffic. Recent studies can achieve high accuracy (e.g., 98%) by leveraging deep neural networks. However, current attacks rely on enormous encrypted traffic data, which are time-consuming to collect. Moreover, large-scale encrypted traffic data also need to be recollected frequently to adjust the changes in the website content. In other words, the bootstrap time for carrying out website fingerprinting is not practical. In this paper, we propose a new method, named Adaptive Fingerprinting, which can derive high attack accuracy over few encrypted traffic by leveraging adversarial domain adaption. With our method, an attacker only needs to collect few traffic rather than large-scale datasets, which makes website fingerprinting more practical in the real world. Our extensive experimental results over multiple datasets show that our method can achieve 89% accuracy over few encrypted traffic in the closed-world setting and 99% precision and 99% recall in the open-world setting. Compared to a recent study (named Triplet Fingerprinting), our method is much more efficient in pre-training time and is more scalable. Moreover, the attack performance of our method can outperform Triplet Fingerprinting in both the closed-world evaluation and open-world evaluation.

In the past years, the practical cross-domain machinery fault diagnosis problems have been attracting growing attention, where the training and testing data are collected from different operating conditions. The recent advances in closed-set domain adaptation have well addressed the basic problem where the fault mode sets are identical in the source and target domains. While some attempts have also been made on the partial and open-set domain adaptations, no prior information of the target-domain fault modes can be usually available in the real industries, that forms a challenging problem in transfer learning. This article proposes a universal domain adaptation method for fault diagnosis, where no explicit assumption is made on the target label set. A hybrid approach with source class-wise and target instance-wise weighting mechanism is proposed for selective adaptation. By using additional outlier identifier, the proposed method can automatically recognize the unknown fault modes while achieving class-level alignments for the shared health states, without knowing the target label set. Experiments on two rotating machine datasets validate the proposed method, which is promising for practical applications under strong data uncertainties.

Data-driven machinery fault diagnosis methods have been successfully developed in the past decades. However, the cross-domain diagnostic problems have not been well addressed, where the training and testing data are collected under different operating conditions. Recently, domain adaptation approaches have been popularly used to bridge this gap, which extract domain-invariant features for diagnostics. Despite the effectiveness, most existing methods assume the label spaces of training and testing data are identical that indicates the fault mode sets are the same in different scenarios. In practice, new fault modes usually occur in testing, which makes the conventional methods focusing on marginal distribution alignment less effective. In order to address this problem, a deep learning-based open-set domain adaptation method is proposed in this study. Adversarial learning is introduced to extract generalized features, and an instance-level weighted mechanism is proposed to reflect the similarities of testing samples with known health states. The unknown fault mode can be effectively identified, and the known states can be also recognized. Entropy minimization scheme is further adopted to improve generalization. Experiments on two practical rotating machinery datasets validate the proposed method. The results suggest the proposed method is promising for open-set domain adaptation problems, which largely enhances the applicability of data-driven approaches in the real industries.

As vast amounts of data are saved, hard drive failure prediction is critical to reducing the cost of data loss and backup. Most existing studies used to detect the anomalous status of a hard drive using self-monitoring, analysis, and reporting technology (SMART) attributes, and then predicted whether the drive would have an impending failure. However, as most researchers focus on a specific model of hard drives, existing studies do not apply to different models due to the model-to-model variations. Furthermore, as anomaly detection algorithms were the focus in the literature, feature selection methods were less compared. This article proposes an evaluation methodology to compare feature selection methods and anomaly detection algorithms for hard drive failure prediction. It can quickly select the optimal algorithms for a specific model of drives. It contains an evaluation mechanism to assess feature selection methods from the perspectives of performance and robustness and assess the performance, the robustness, the efficiency, and the generalization of anomaly detection algorithms. Experiments on two data sets are implemented for validation, and the proposed method can achieve better performance than the existing approaches in the literature.

Rotating machinery fault diagnosis problems have been well-addressed when sufficient supervised data of the tested machine are available using the latest data-driven methods. However, it is still challenging to develop effective diagnostic method with insufficient training data, which is highly demanded in real-industrial scenarios, since high-quality data are usually difficult and expensive to collect. Considering the underlying similarities of rotating machines, data mining on different but related equipments potentially benefit the diagnostic performance on the target machine. Therefore, a novel transfer learning method for diagnostics based on deep learning is proposed in this article, where the diagnostic knowledge learned from sufficient supervised data of multiple rotating machines is transferred to the target equipment with domain adversarial training. Different from the existing studies, a more generalized transfer learning problem with different label spaces of domains is investigated, and different fault severities are also considered in fault diagnostics. The experimental results on four datasets validate the effectiveness of the proposed method, and show it is feasible and promising to explore different datasets to improve diagnostic performance.

This paper proposes a multiobjective optimization method for signal control design at intersections in urban traffic network. The cell transmission model is employed for macroscopic simulation of the traffic. Additional rules are introduced to model different route choices from origins to destinations. Vehicle turning, merging, and diverging behaviors at intersections are considered. A multiobjective optimization problem (MOP) is formulated considering four measures in network traffic performance, i.e., maximizing system throughputs, minimizing traveling delays, enhancing traffic safety, and avoiding spillovers. The design parameters for an intersection include turning signal type, cycle time, signal offset, and green time in each phase. The resulting high-dimensional MOP is solved with the genetic algorithm (GA). An algorithm is proposed to assist the user to select and implement the optimal designs from the Pareto optimal solution set. A case study in a grid network of nine intersections is carried out to test the optimization algorithm. It is observed that the proposed method is able to achieve the optimal network performance with different traffic demands. The convergence and coefficient selection of GA are discussed. The guidelines for network signal design and operation from the current studies are presented.