It is well known that data drift may occur in complex industrial processes, which can cause changes in the distribution of data sampled by sensors. Therefore, the ability to generalize across unseen domains is essential for monitoring systems deployed in industrial processes. The invariant-complement domain generalization (ICDG) is proposed to alleviate data drift in industrial processes. This study reveals how the proposed covariant representation complements the invariant representation. Additionally, it derives novel theoretical error bounds characterizing the relationship between seen and unseen domains. Intuitively, the proposed invariant-complement representation method mitigates the influence of variant factors and encourages the learning of invariant and covariant representations. From the perspective of information theory, the boundaries for invariant and covariant representations are established and integrated as a joint learning objective with multiple information constraints. Theoretically, we elucidate that optimizing the ICDG objective function is equivalent to minimizing the upper bound of the empirical risk associated with unseen domains. This result helps ensure accurate prediction of quality variables under data drift. Case studies on the gas turbine (GT) dataset and the actual polyester esterification dataset validate the effectiveness of the proposed ICDG. The code is available athttps://github.com/heheding/ICDG

Different sampling rates are common for different variables in industrial processes because of the different electrical properties and requirements of sensors. Especially the sampling rate of quality variables is significantly lower than that of process variables. However, most soft sensors assume that industrial data is uniformly sampled, which differs significantly from actual industrial systems and may affect decision-making in the production process. An interpretable disentangled transfer learning (IDTL) quality prediction is proposed suitable for multirate industrial processes. First, a setness constructor is designed to diversify the original multirate data into multiple multirate sets to preserve information without data loss. Then, a disentangled transfer learning (TL) approach is proposed to infer domain-invariant and domain-specific representations from multiple multirate sets, thereby revealing the intrinsic properties of multirate industrial processes and improving the soft sensor performance. From the perspective of information theory, the theoretical representations for disentanglement and their connection to TL are established, laying a solid theoretical foundation for subsequent TL under complex working conditions. Our theoretical analysis shows that interpretable disentangled TL (IDTL) achieves optimal disentangled representations in equilibrium. Case studies of the debutanizer column dataset and the actual polyester esterification dataset validate the effectiveness of the proposed IDTL. Code is available at https://github.com/heheding/IDTL.

Most fine-grained visual recognition methods endeavor to directly locate discriminative regions in intricate environments, but tend to overlook the object’s holistic structure, which may lead to misclassification due to overemphasizing incorrect areas. In this paper, we propose a coarse-to-fine paradigm, which prioritizes locating holistic structural regions of the target object, followed by a gradual search to locate discriminative areas. Specifically, we first design the “look into object” module to locate the areas encompassing the target’s holistic structure using prior information. Subsequently, without introducing additional parameters, we design a partial focus searching module to enhance feature representations of discriminative regions within the target’s structural composition. Ultimately, we segregate the foreground components from the original image, attaining a more precise characterization of the target. Furthermore, we demonstrate the practical application potential of our model in real-world industries through our self-constructed DHU-Fine-grained-6000 dataset. Comparative experiments on three public datasets indicate that the superiority of our approach over many recent methods and holds promising application potential in industrial production processes.

This article presents a robust variational Bayesian (VB) algorithm for identifying piecewise autoregressive exogenous (PWARX) systems with time-varying time-delays. To alleviate the adverse effects caused by outliers, the probability distribution of noise is taken to follow a t -distribution. Meanwhile, a solution strategy for more accurately classifying undecidable data points is proposed, and the hyperplanes used to split data are determined by a support vector machine (SVM). In addition, maximum-likelihood estimation (MLE) is adopted to re-estimate the unknown parameters through the classification results. The time-delay is regarded as a hidden variable and identified through the VB algorithm. The effectiveness of the proposed algorithm is illustrated by two simulation examples.

In this article, an approach for identification of an errors-in-variable system whose output is contaminated by heteroscedastic noise is developed. A Markov chain is applied to depict the correlation of the switching of heteroscedastic noise model. The estimation of model parameters adopts a variational Bayesian algorithm. The advantage of the Bayesian approach is the full probability description of the estimates while the classical expectation-maximization algorithm only provides point estimation. A simulated numerical example and an experimental study on a polyester fiber process are provided to demonstrate the effectiveness of the proposed method. Three performance indexes, normalized mean-absolute error, mean-relative error and root-mean-squared error, are used to evaluate the performance of the proposed algorithm. Meanwhile, Monte Carlo cross validations are performed to demonstrate the effectiveness and superiority of the proposed algorithm.

A service optimization method for polyester fiber production process is proposed. According to the production batch and production specifications, the method considers the service cost as the optimization objective, and uses data model to determine the specific process parameters in the polyester fiber production process. First, two options for the overall process of polyester fiber are introduced: on-demand manufacturing and product development. Second, the impact of different batch request tasks on the performance index of each stage is determined. Finally, the service optimization measures of different batches are proposed. By comparing the similarity between the current data samples and the overall data, the optimal production plan of the overall production process is formed. Simulation results show that the immune algorithm inspired from endocrine regulation has the best performance on the optimal decision-making combination, which is helpful for the development of new polyester products. We investigate how to reduce energy consumption of system resources, and how to choose the best service from a large number of candidate services. In the overall polyester fiber production process, users are not only consumers, but also designers and producers, achieving the real “integration of production and consumption”.

Autoencoder (AE) is a deep neural network that has been widely utilized in process industry owing to its superior abilities of feature extraction and data reconstruction. Recently, assuming the latent variables to be random variables, a probabilistic variant of it called variational autoencoder (VAE) has achieved a major success in different applications. In this article, we develop two novel submodels based on deep VAEs (DVAE), which are further utilized to establish a soft sensor framework. By the use of our first submodel known as supervised DVAE (SDVAE), the distribution information of latent features can be obtained. This is used as a prior of the second submodel known as the modified unsupervised DVAE (MUDVAE). Then, a new soft sensor framework can be constructed by combing the encoder of SDVAE with the decoder of MUDVAE. Since our designed VAE has superior ability in data reconstruction, it also works well under the missing data situation which is common in process industries due to sensor failures. Thus, we extend the proposed soft sensor framework to handle the missing data situation. The effectiveness of our proposed soft sensor frameworks is finally demonstrated via an industrial polymerization dataset.

In this paper, a dynamic leader selection process of a multirobot system with leader-follower strategies is studied in terms of formation control. A fuzzy inference system is employed to evaluate the status of robots by means of their states. Based on the status, an affection-based model is proposed to trigger a leader selection module. Followers send out unsatisfied signals when they are disappointed at the current leader. The abashment value of the leader changes with its own status as well as the number of unsatisfied signals received from its followers. When its abashment value goes beyond a given threshold, a leader reselection process is triggered. Moreover, a swap-greedy algorithm is proposed to approximate the optimal solution for confirming the leader-follower relationship, which can be described as a combinatorial optimization problem to minimize the total travel distance of all the robots. Extensive simulation results demonstrate that the proposed model can improve the probability of a robot team escaping from local extreme points significantly, and even in the case of leader failure, the team can reselect a leader autonomously and keep moving toward the target.