Showing papers in "Journal of Process Control in 2020"

TL;DR: A new DNN model, one-dimensional convolutional auto-encoder (1D-CAE) is proposed for fault detection and diagnosis of multivariate processes in this paper, and provides an effective platform for deep-learning-based process fault detectionand diagnosis ofMultivariate processes.

TL;DR: A stacked supervised auto-encoder is proposed to pretrain the deep network and obtain deep fault-relevant features from raw input data and is tested on the Tennessee–Eastman benchmark process and a real industrial hydrocracking process, showing the effectiveness and flexibility of SSAE.

TL;DR: A systematic review of various state-of-the-art kernel-based methods with applications for data pre-processing, sample selection, variable selection, model construction and reliability analysis of soft-sensors.

TL;DR: The proposed physics-based RNN models are utilized in model predictive controllers and applied to a chemical process network example to demonstrate their improved approximation performance compared to the fully-connected RNN model that is developed as a black box model.

TL;DR: To account for the highly complex dynamics of industrial process and additional requirements imposed by smart and optimal manufacturing systems, an extended state space descriptive system is designed and, based on the descriptive system, a hybrid first principles/machine learning modeling framework is proposed.

TL;DR: The proposed fault diagnosis model based on the optimized long short-term memory (LSTM) network has better performance in chemical process fault diagnosis than the BP neural network, the multi-layer perceptron method and the original LSTM network.

TL;DR: The VA-WGAN combining VAE with Wasserstein generative adversarial networks (WGAN) as a generative model is established to produce new samples for soft sensors by using the decoder of VAE as the generator in WGAN.

TL;DR: This study develops a model-based RL method, which iteratively learns the solution to the HJB and its associated equations, and shows that the use of DNNs can significantly improve the performance of a learned policy in the presence of uncertain initial state and state noise.

TL;DR: In this paper, the authors present the methodology and field test demonstration of a computationally efficient implementation of the white-box MPC in an office building in Belgium, where the detailed model of the building is based on first-principle physical equations.

TL;DR: In this paper, a Dynamic Sliding Mode Controller (DSMC) based on the Iinoya and Altpeter approach and the SLMC design procedure is proposed for chemical processes of high order with long dead time, and with inverse response.

TL;DR: Experimental results of two real industrial processes show that the adaptive monitoring strategy based on recursive CA can effectively adapt to normal process changes without frequent model updating.

TL;DR: A new DNN, manifold regularized stacked autoencoders (MRSAE) for fault detection in complex industrial processes is proposed and the comparison between MRSAE and other typical DNNs on a complex numerical process and two benchmark processes indicates the effectiveness of the proposed method.

TL;DR: The performance, advantages, and disadvantages of the two strategies are analyzed using demand-response scenarios with varying levels of fluctuations in electricity prices, as well as considering the cases of known, instantaneous, and completely unknown load changes.

TL;DR: The proposed streaming parallel VBSFA (SP-VBSFA) algorithm not only relieves the computing pressure of modeling big process data, but also improves the prediction accuracy and further reduces the tracking time delay for process variations.

TL;DR: This paper gives an alternative approach to the existing rainfall/runoff linear and nonlinear models by the utilization of a hybrid system consisting in a Piecewise Auto-Regressive eXogeneous (PWARX) structure identified using an approach that alternates between data assignment and parameter estimation.

TL;DR: This work proposes a two-step procedure, which outperforms the conditional GC test and provides an easy way to identify the root cause of process disturbances, and is illustrated with the application to the Tennessee Eastman benchmark process.

TL;DR: A multi-feature extraction technique based on PCA is proposed for nonlinear dynamic process monitoring that integrates dynamic inner PCA, PCA and kernel PCA methods through a serial structure to extract the dynamic, linear and nonlinear features among the process data.

TL;DR: The manuscript shows that in the case of highly nonlinear dynamics, as observed in the proppant concentration, use of canonical functions in the observable basis fails and in such cases, a priori system knowledge can be leveraged to choose the required basis.

TL;DR: A novel algorithm named information concentrated variational auto-encoder (IFCVAE) is proposed, which aims to extract the latent variables which represent both process information and quality information from the quality-related and unrelated aspects of industrial processes.

TL;DR: Various nonlinear latent variable models based on autoencoder (AE) are developed in order to extract deeper nonlinear features from process data and provide a deep generative structure for nonlinear process monitoring and quality prediction.

TL;DR: An output- relevant VAE is proposed for extracting output-relevant features from high-dimensional dataset using correlation between process variables to establish a model between the input and the corresponding output at the query sample.

TL;DR: A data augmentation method, which is based on Generative Adversarial Networks and aided by Gaussian Discriminant Analysis, for enhancement of fault classification accuracy and deployed and parallelly trained on Tensorflow platform, suitable for applications likeData augmentation and imbalanced fault classification in industrial big data environments.

TL;DR: It is demonstrated that deterministic MPC fails to properly capture disturbances and this translates into economic penalties associated with peak demand charges and constraint violations in thermal storage capacity (overflow and/or depletion).

TL;DR: An augmented subcutaneous model of type 1 diabetic patients (T1DP) is proposed first by estimating the model parameters with the aid of nonlinear least square method using the physiological data and a nonlinear adaptive controller is proposed to tackle two important issues of intra-patient variability and uncertain meal disturbance.

TL;DR: A new data-driven algorithm called enhanced canonical variate analysis with slow feature (ECVAS) and corresponding monitoring strategy are proposed for dynamic process monitoring, which achieves in-depth understanding of process dynamics under control actions and helps identify normal changes in operating conditions.

TL;DR: A novel Laplacian regularized robust principal component analysis (LRPCA) framework is proposed, where the “robust” comes from the introduction of a sparse term, and not only can represent the global low-dimensional structures, but also capture the intrinsic non-linear geometric information.

TL;DR: It is found that the proposed ANN initialization strategy mostly results in the same control policy as the shift-initialization strategy, and the proposed QP-based method yields a good compromise between finding the optimal control policy and solution time.

TL;DR: This commonly overlooked problem in data-driven soft sensor modeling is illustrated and solved and the proposed VTR-based model can effectively learn the VTD values, which can reconstruct and recover the original data pattern and thus significantly help increase the generalization performance of soft sensor models.

TL;DR: A significantly better performance is achieved in terms of classical indexes and applicability in the in-silico patients from the FDA-accepted UVA/Padova simulation platform, considering the most challenging scenarios.

TL;DR: A novel incipient fault detection method based on robust support vector data description (RSVDD) is proposed and on the basis of traditional SVDD, both normal samples and faulty samples are introduced for RSVDD modeling and the computation of sphere radius is improved.