Accurate forecasting of household energy load and generation is critical for energy management systems, especially in the context of the rapid development of smart grids and renewable energy. However, privacy concerns often arise when handling sensitive household energy data. Federated learning (FL), as a privacy-preserving distributed learning method, enables collaborative model training across households without exposing sensitive energy data. Nevertheless, traditional federated learning still faces challenges in meeting the personalized needs of households due to the differences in consumption patterns among households, which affects the forecasting accuracy. In this paper, we propose a novel personalized FL method that combines mixture of expert and conformal predictions to improve forecasting accuracy while quantifying prediction uncertainty. Our method utilizes personalized federated learning (PFL) to develop a personalized model for each household that captures its unique consumption behavior. The mixture of experts dynamically integrates global and local personalized models to enhance prediction and adaptability. In addition, uncertainty quantification is achieved through conformal prediction, providing reliable prediction interval. Experiments conducted on two real-world household energy datasets demonstrate that our method outperforms existing approaches in terms of prediction accuracy and uncertainty assessment.

Deep learning has become increasingly important in fault diagnosis, but it relies on a large amount of high-quality labeled data. Collecting data from distributed machines can expand the dataset, but it usually leads to privacy concerns. Moreover, since the operating conditions are complex in real-world applications, the collected training data and the test data often have different distributions. Therefore, a well-trained model on the training data may not be suitable for test data due to the domain shift. To preserve privacy and to mitigate the domain shift, in existing federated transfer learning fault diagnosis methods, distributed machines exchange model parameters and features rather than raw data with the central server. However, such methods suffer from a single point of failure and high communication burden. To address these issues, we propose a fully decentralized federated transfer learning fault diagnosis method. More specifically, the proposed method obtains a pretrained model among source nodes with labeled training data where each source node exchanges model parameters with its neighboring source nodes. Moreover, a novel transfer learning strategy is proposed, which aligns features of test data at the target node with features of training data at its connected source nodes to mitigate misclassifications resulting from the domain shift. The effectiveness of the proposed method is verified by various experiments on two public bearing datasets.

The successive switching nature of multimode processes, coupled with data scarcity, challenges traditional quality prediction models. Specifically, the difficulty of simultaneously collecting abundant labeled datasets from all modes forces the model to update its parameters as modes switch. This leads to the forgetting of historical mode knowledge and hinders the aggregation of knowledge, thereby degrading generalization across modes. To this end, we propose a novel continual semisupervised graph echo state network ($\text{CS}^{2}$GESN). First, a semisupervised graph echo state network ($\text{S}^{2}$GESN) is designed based on the graph smoothing assumption to extract dynamic information from unlabeled samples within each mode. The$\text{S}^{2}$GESN model then evolves into a continual model,$\text{CS}^{2}$GESN, employing an elastic weight consolidation strategy for parameter importance estimation derived from pseudoinverse parameter optimization, facilitating the accumulation of historically learned knowledge. This manner alleviates performance deterioration from data scarcity and information forgetting, and enables more flexible modeling of successive arriving operating modes. The superiority and feasibility of the proposed method are demonstrated through its application to the Tennessee Eastman process and the three-phase flow facility process.

Conventional data-driven dynamic process monitoring methods usually rely on data collected at a single sampling rate. The effectiveness of these approaches typically diminishes when analyzing data from multiple sampling rates. To address this gap, this article introduces a new total multirate linear Gaussian state space model. This model is designed for modeling and monitoring in dynamic processes that involve data from various sampling rates. It works by establishing global dynamic latent variables that span across process variables and extracting local static latent variables for each sampling rate. For effective fault detection at different sampling rates, the model incorporates three kinds of statistics. The effectiveness of the proposed method in process monitoring is validated using the multiphase flow facility benchmark and a real papermaking wastewater treatment process.

In many real-world industrial scenarios, health assessment of bearings with no prior faults (BNFs) is hindered by label-insufficient and sample-insufficient problem. To address this issue, this article proposes a novel normal-faulty adversarial bridging framework (NFABF) to unify three types of samples—normal, faulty, and suspicious—from multibearing models into a common latent space, thereby facilitating real-time BNF health assessment. Specifically, an adversarial bridging autoencoder is devised to simultaneously reconstruct normal samples and “deconstruct” faulty samples, while employing an information-entropy (IE)–based method to refine pseudolabels for suspicious samples. This process enables the feature encoder to effectively embed suspicious samples into the transition region between normal and faulty features. Furthermore, a bridging alignment strategy, integrating both maximum mean discrepancy and feature cohesion loss, is introduced to reduce discrepancies among the three sample types, while a temporal continuity constraint enforces a realistic degradation evolution. Last, a health index is constructed by combining reconstruction error and fault probability, and a regression model is incorporated to estimate the remaining useful life. The proposed framework is validated using real main shaft bearing data from wind turbine systems and extensively compared with multiple baselines. Experimental results demonstrate that the NFABF achieves superior performance in bearing health assessment for the BNF scenario.

Traditional process monitoring methods often rely on data with uniform sampling rates, which may lead to the loss of valuable information across both time and space dimensions. Moreover, multirate data exhibits strong autocorrelation and cross-correlation among various sampling rates. Effectively capturing these characteristics is crucial for accurately monitoring process variations. In this article, a total structure multirate autoregressive dynamic latent variable (Ts-MARDLV) model is proposed, which establishes global dynamic latent variables for all measurements and local static latent variables for each sampling rate, effectively analyzing the autocorrelation and cross-correlation of samples. For multirate process monitoring, the Ts-MARDLV model-based fault detection schemes are developed. Three diverse fault detection statistical metrics are constructed to monitor faults in different latent spaces. The proposed method is validated on multiphase flow datasets and a real papermaking wastewater process, demonstrating its superior effectiveness compared to single or multirate methods.

Quality prediction is beneficial to intelligent inspection, advanced process control, operation optimization, and product quality improvements of complex industrial processes. Most of the existing work obeys the assumption that training samples and testing samples follow similar data distributions. The assumption is, however, not true for practical multimode processes with dynamics. In practice, traditional approaches mostly establish a prediction model using the samples from the principal operating mode (POM) with abundant samples. The model is inapplicable to other modes with a few samples. In view of this, this article will propose a novel dynamic latent variable (DLV)-based transfer learning approach, called transfer DLV regression (TDLVR), for quality prediction of multimode processes with dynamics. The proposed TDLVR can not only derive the dynamics between process variables and quality variables in the POM but also extract the co-dynamic variations among process variables between the POM and the new mode. This can effectively overcome data marginal distribution discrepancy and enrich the information of the new mode. To make full use of the available labeled samples from the new mode, an error compensation mechanism is incorporated into the established TDLVR, termed compensated TDLVR (CTDLVR), to adapt to the conditional distribution discrepancy. Empirical studies show the efficacy of the proposed TDLVR and CTDLVR methods in several case studies, including numerical simulation examples and two real-industrial process examples.

A critical side effect of online social networks’ flourishing is fast-spreading rumors on the Internet, making the information diffusion control on social networks a fundamental requirement. While information diffusion control has received extensive attention at the global level, there have been fewer user-level diffusion control studies under minimal budget. In this article, we study the information diffusion at the user level and propose a diffusion control method based on gradient information to generate an optimized network structure, namely ConTroL information DIFFusion (CTL-DIFF). CTL-DIFF targets a user through subtle modifications of its local network structure. It first selects the edges with the largest absolute gradient based on the prediction model to optimize the original network’s structure. It then employs several prediction methods to verify whether the target user’s social action status is controlled. CTL-DIFF achieves state-of-the-art control performance with a minimum budget, comparing with five baselines based on edge centrality strategies on four real-world datasets. We extend the diffusion control from user-level to global-level, comparing with four baselines on three datasets. Experimental results show that CTL-DIFF can effectively control information diffusion in the global social network by identifying and controlling the most influential users.

Infrared thermography is an economical nondestructive testing technique for structural health monitoring of composite materials. However, the nonlinear nature of the thermographic data and the adverse effects of noise and inhomogeneous backgrounds prevent it from achieving satisfactory results. Most of the existing thermographic data analysis methods are supervised and/or linear, which, therefore, are not favorable for nonlinear feature extraction of unlabeled thermograms. In this article, a deep autoencoder thermography (DAT) method is proposed for detecting subsurface defects in composite materials. The multilayer network structure of DAT can handle nonlinear temperature profiles, and the output of the intermediate hidden layer is visualized to highlight defects. The layer-by-layer feature visualization reveals how the model extracts defect features. A loss inflection point scheme is utilized to determine a suitable depth of the model. Moreover, a new quantitative index is proposed to compare the defect detectability of different methods.

Accurate depiction of the process characteristics of dynamic multiphase flows using a data-driven model is a challenge in industrial practices. Collection of sufficient data is costly and cumbersome, and it is difficult to identify representative data efficiently. This article develops an active learning method to explore information from multiphase flow process data, thus facilitating smart process modeling and prediction. An index is proposed to describe the process dynamics and nonlinearity using a probabilistic model, facilitating determination of informative data. The subsequent absorption of these data into the training set enhances the model quality gradually. This is relevant especially for transitional regions exhibiting dynamic information. In addition, a simple and efficient criterion to judge the learning termination has been designed. Consequently, new representative data are explored and learned in a sequential manner. The experimental results of two industrial multiphase flows demonstrate the advantages of the proposed method.

Software defect prediction aims to automatically locate defective code modules to better focus testing resources and human effort. Typically, software defect prediction pipelines are comprised of two parts: the first extracts program features, like abstract syntax trees, by using external tools, and the second applies machine learning-based classification models to those features in order to predict defective modules. Since such approaches depend on specific feature extraction tools, machine learning classifiers have to be custom-tailored to effectively build most accurate models.

Increasing deep neural networks are applied in solving graph evolved tasks, such as node classification and link prediction. However, the vulnerability of deep models can be revealed using carefully crafted adversarial examples generated by various adversarial attack methods. To explore this security problem, we define the link prediction adversarial attack problem and put forward a novel iterative gradient attack (IGA) strategy using the gradient information in the trained graph autoencoder (GAE) model. Not surprisingly, GAE can be fooled by an adversarial graph with a few links perturbed on the clean one. The results on comprehensive experiments of different real-world graphs indicate that most deep models and even the state-of-the-art link prediction algorithms cannot escape the adversarial attack, such as GAE. We can benefit the attack as an efficient privacy protection tool from the link prediction of unknown violations. On the other hand, the adversarial attack is a robust evaluation metric for current link prediction algorithms of their defensibility.

The subspace learning (dimensionality reduction) algorithms have played an important role in the analysis of thermographic data: a key step in infrared thermography-based nondestructive testing of subsurface defects in composite materials. However, one of its branches, manifold learning, with excellent ability to preserve local data structure, is rarely applied. In this article, a spatial-neighborhood manifold learning (SNML) framework is proposed for thermographic data analysis. Different from traditional manifold learning methods, SNML uses the spatial-neighborhood information instead of the traditional k-nearest neighbors, or ε-neighborhood, to construct the adjacency graph. This overcomes the difficulty of parameter selection and extracts local features in images in a more reasonable way. Additionally, the data preprocessing step and the means of thermographic data normalization in the proposed framework are discussed. For performance comparison, three traditional manifold learning methods are also implemented. The experiments on carbon fiber-reinforced polymer specimens demonstrate the validity and feasibility of SNML.

Nondestructive testing (NDT) is an economical way of detecting subsurface defects in composite products. Infrared thermography serves as a popular NDT method due to its high efficiency and low cost. However, defect identification by directly visualizing thermal images is difficult owing to the nonuniform background and noise. Recently, data analysis methods have been introduced to thermal image processing, including principal component analysis, which is known for its good performance in dimensionality reduction, feature extraction, and noise reduction. However, most of these methods can only extract linear features. In this article, a multiview learning-based autoencoder, which can process not only nonlinear features but also sequential attributes, is utilized in thermographic data analysis. After extracting the low-dimensional features by multiview learning, a background elimination step is conducted to highlight the locations and shapes of the defects. The experimental results demonstrate the feasibility of the proposed method.