Complex neural networks with deep structures are beneficial for solving problems such as fault classification and health prediction of industrial equipment due to their powerful feature extraction capabilities. Unfortunately, corresponding complex models designed based on deep learning algorithms require huge computational and memory resources, making them difficult to achieve effective edge deployment. In order to solve this difficulty with practical industrial significance, this paper proposes an online knowledge distillation framework for machine health prognosis. Within this framework, the learned knowledge of complex networks can be distilled to simple networks that can be deployed on edge devices in sites. Specifically, the response-based knowledge distillation module, feature-based knowledge distillation module, and relation-based knowledge distillation module are respectively designed to achieve effective information transmission from different levels. Furthermore, the inherent differences between simple and complex networks have been fully considered for their impact on the efficiency of knowledge distillation, and an adaptive mutual learning strategy has been contrapuntally proposed to address this limitation. Multiple online knowledge distillation experiments were conducted on two different sets of run-to-failure datasets of mechanical key components with different pairs of complex and simple networks to verify the effectiveness of the proposed framework. The experimental results show that the simple student-networks can effectively improve prediction performance after receiving knowledge distillation from the complex teacher-networks, providing a new solution for machine health prognosis under the premise of edge deployment.

The Industrial Internet of Things (IIoT) greatly facilitates prognostics and health management of complex industrial systems, wherein the vast amount of real-time data from the IIoT improves intelligent predictive maintenance of industrial systems. When processing industrial IoT data across devices, traditional subdomain adaptation-based methods ignore the local similarities across domains. Also, if fault classes are used to define subdomains, these methods may not be applicable when the target domain is unlabeled or has limited labels. To address the above challenges, a Graph-embedded Subdomain Adaptation Network (GSAN)-based approach is proposed to predict the remaining useful life under different machines in IIoT. Specifically, a manifold subdomain representation is established by manifold learning and local manifold discrepancies between each pair of manifold subdomains with the highest similarity are minimized. To maintain a divisible margin for each manifold, a self-supervised intra-manifold regularization module is developed. An extensive evaluation of six transfer scenarios is performed, and the experimental results show that GSAN can achieve more significant outcomes. This can provide some guidance for future work on prognostics across devices and subdomains.

Fault detection or anomaly detection relies heavily on learning from datasets where only normal samples are available, resulting in the emergence of numerous one-class classification (OCC) methods. However, learning discriminative deep representatives with good generalization from cross-domain positive samples remains challenging. Therefore, this work proposes an end-to-end framework, deep transfer one-class classification (DTOCC) for unsupervised fault detection, which combines adversarial generative OCC and distribution alignment from the perspective of manifold learning. Specifically, pseudo-negative samples are generated outside the positive manifold, facilitating the model to learn discrimination with respect to normal and anomaly. Further, cross-domain positive samples are aligned in log-Euclidean manifold space to enhance representation learning. Then, we provide the specific implementations for fault detection and validate its superiority through case studies on multiclass and run-to-failure datasets, simulating both offline and online scenarios.

Deep neural networks have promoted the technology development of fault classification and remaining useful life (RUL) prediction for mechanical equipment due to their powerful nonlinear feature extraction capability. However, the performance of traditional deep learning models is limited by the depth of networks, which is directly related to the training consumption. In addition, the parameters of networks can only be updated by retraining when faced with newly acquired data. To address the above problems, an incremental learning method based on a temporal cascade broad learning system (TCBLS) is proposed for the RUL prediction of machinery with newly acquired data. Specifically, linear and nonlinear feature information is first learned by the TCBLS. The ridge regression method is developed to calculate the weights of the network and establish an end-to-end mapping between the feature information layer and the prediction layer. Finally, the incremental learning of new data and the incremental learning of nodes are proposed for adaptively updating the weights of the network in the face of newly acquired data and insufficient prediction accuracy. The effectiveness of the proposed method is verified by four run-to-failure datasets. The comparison results with classical deep learning models show that the proposed method is promising for RUL prediction as it achieves high prediction accuracy while saving training time consumption across orders of magnitude and effectively handling newly acquired data without retraining.

The industrial gearboxes usually work in harsh and variable conditions, which results in partial failure of gears or bearings. Accordingly, the continuous irregular fluctuations of gearbox under variable conditions maybe increase the intraclass difference and reduce the interclass difference for the monitored samples. To this end, a new intelligent fault diagnosis method of gearbox based on adaptive intraclass and interclass convolutional neural network (AIICNN) under variable working conditions is proposed. The core of the proposed algorithm is to apply the designed intraclass and interclass constraints to improve the distribution differences of samples. Meanwhile, the adaptive activation function is added into the 1-D convolutional neural network (1dCNN) to enlarge the heterogeneous distance and narrow the homogeneous distance of samples. Specifically, the training sample subset with intraclass and interclass spacing fluctuations under variable conditions is first converted into frequency domain through the fast Fourier transform (FFT), and the designed AIICNN algorithm is employed for model training. Afterward, the testing subset is provided to the trained AIICNN algorithm for fault diagnosis. The experimental data of the planetary gearbox test rig verify the feasibility of the proposed diagnosis method and algorithm. Compared with other methods, this method can eliminate the difference of sample distribution under variable conditions and improve its diagnostic generalization.

Advanced mechatronics equipment requires reliable and effective performance degradation assessment to guarantee long-term operations. Current data-driven predictions endow the operation and maintenance of mechatronic equipment flexibly and intelligently. However, the sufficient and labeled data in real industrial scenes may not be satisfied, resulting in negative impacts of overfitting and time-consuming annotations. In this article, we propose a novel prognostic model, namely unsupervised meta gated recurrent unit (UMGRU) containing a dual-cycle learning architecture with the designed clustering assignment module to deal with few-shot prognostics under unlabeled historical data. It integrates the strength of double gradient based optimizations for abstracting general degradation knowledge and offering a sensitive model status for precisely online adaptation with limited on-site data. Besides, mini-batch pseudolabels are automatically assigned within each inner cycle learning and further participate in parameter upgrades. Finally, both experimental and industrial data are used to verify the effectiveness of UMGRU.

This article develops a novel collaborative health monitoring framework based on hierarchical symbol transition entropy (HSTE) and 2-D-extreme learning machine (2-D-ELM) without fusion. In the proposed framework, a novel metric called symbol transition entropy (STE) is first presented to evaluate the dynamical complexity of time series through multistep transition and the joint probability distribution of the symbol state and its transition state. Compared with the existing entropy algorithms, STE has better robustness and captures more detailed dynamical changes. Subsequently, a new feature representation method called HSTE is proposed by combining STE with the hierarchical analysis. The two-order tensor features can be constructed for multichannel data by stacking HSTE values extracted from each single-channel data. Finally, 2-D-ELM is incorporated to identify the extracted two-order tensor features without vectorization. The feasibility of the proposed schemes is verified through simulation and experimental studies, and the final results confirm that the developed schemes have better performance than the existing entropy-based collaborative fault diagnosis methods.

The labeled monitoring data collected from the electromechanical system is limited in the real industries; traditional intelligent fault diagnosis methods cannot achieve satisfactory accurate diagnosis results. To deal with this problem, an intelligent fault diagnosis method for electromechanical system based on a new semisupervised graph convolution deep belief network algorithm is proposed in this article. Specifically, the labeled and unlabeled samples are first employed to design a new adaptive local graph learning method for constructing the graph neighbor relationship. Meanwhile, the labeled samples are applied to describe the discriminative structure information of data via the latest circle loss. Finally, the local and discriminative objective functions are reconstructed under the semisupervised learning framework. The experimental results from the motor-bearing system demonstrate that the method can achieve 98.66 % accuracy with only 10 % of training labeled data, which indicates that it is a promising semisupervised intelligent fault diagnosis method.

Health evaluation is attracting more and more attention in the domain of machinery prognostic and health management (PHM). Meanwhile, few studies have been devoted to health evaluation under variable working conditions and few shots learning, which are common situations under industrial sites. Thus, this shortcoming becomes the motivation of our study. We propose subspace meta-learning (SML) that integrates the strengths of knowledge transfer, constructing the statistically relevant latent subspace, and meta learning, realizing few shots prognostics. To be specifically, time-frequency images are first extracted with sliding windows along with the vibration signals across different life experiments of rolling bearings. Then, two-dimensional domain adaptation based on high order statistical properties is utilized to construct latent subspace and generate meta degradation knowledge. Finally, the convolutional layer based meta learning under model-agnostic learning mode is set up based on the time-frequency degradation knowledge. For a transparent test of our proposed SML health evaluation methodologies, public FEMTO-ST bearing datasets are employed for verifications, and comparisons are also conducted between existing prediction methods. Prediction performances reveal that the superiority of SML under few-shot prognostics.

This paper proposes a new sensor deployment strategy based on quantitative cause-effect graph (QCEG) to handle the heterogeneity among the properties of sensors and faults. A QCEG is developed to model the cause-effect relationship between the system faults and sensor readings. A multi-objective optimization is performed to facilitate the monitoring of single-station multistep manufacturing process (SMMP). A stream of fault information model is built to describe the propagation of fault state in the SMMP. By means of state-space transformation, a detection factor is used to provide the initial sensor deployment. The optimal sensor deployment in an SMMP is achieved by an improved shuffled frog leaping algorithm (ISFLA), which minimizes the fault unobservability, maximizes the system stability, and minimizes the cost for the whole system, under the constraints on detectability, stationarity, and limited resources. Two experimental investigations on an assembly unit and a manufacturing unit are conducted to verify the methodology. Comparative studies demonstrate that the proposed QCEG is able to overcome the shortcomings of directed graph (DG) in handling sensor heterogeneity and multiple objectives. As a goal-oriented swarm-intelligence search strategy, the ISFLA performs better than the popular integer programming in dealing with the multi-objective optimization problem.

This study proposes a hybrid model for evaluating surface roughness in hard turning using a Bayesian inference-based hidden Markov model and least squares support vector machine (HMM-SVM). The model inputs are multidirectional fusion features that are extracted from the acquired monitoring signals through independent component analysis and singular spectrum analysis. Based on a detailed analysis of the workpiece surface formation mechanism, the cutting vibration signals are determined as monitoring signals and an experimental scheme based on the multifeed rate is designed. The error rate of HMM-SVM is further reduced by introducing the stratification factor comparison method rather than using the conventional probability comparison method. A five-step iterative algorithm is presented to select and optimize the training set, which effectively solves the problems of precision degradation and training data insufficiency. Experimental studies show that the proposed model can accurately predict the surface roughness in case of missing samples. The advantages of the proposed model over least squares support vector machine (LSSVM) and multiple regression approaches are demonstrated via statistical analysis. Note to Practitioners-As an alternative to traditional grinding, hard turning is an attractive machining method, in which surface quality is a crucial measurement index. However, under the scenario of sample missing, a straightforward and relatively accurate model for predicting surface roughness is challenging to establish using conventional strategies. This paper reports on a new HMM-SVM model based on Bayesian inference for modeling and predicting surface roughness in hard turning. The samples are classified based on the accuracy grade of surface roughness according to the GB/T1031-2009 standard using the expectation maximization algorithm and HMM, which is superior in small-sample classification problem. LSSVM is employed to estimate surface roughness. The effectiveness of the proposed model is demonstrated through experimental investigations. The reported methodology can also be extended to other related fields.