Search

In this paper we introduce the current thrust of development in machine learning and artificial intelligence, fueled by advances in statistical learning theory over the last 20 years and commercial successes by leading big data companies. Then we discuss the characteristics of process manufacturing systems and briefly review the data analytics research and development in the last three decades. We give three attributes for process data analytics to make machine learning techniques applicable in the process industries. Next we provide a perspective on the currently active topics in machine learning that could be opportunities for process data analytics research and development. Finally we address the importance of a data analytics culture. Issues discussed range from technology development to workforce education and from government initiatives to curriculum enhancement. [ABSTRACT FROM AUTHOR]

Deep learning models such as the long short-term memory (LSTM) network have been applied for dynamic inferential modeling. However, many studies apply LSTM as a black-box approach without examining the necessity and usefulness of the internal LSTM gates for inferential modeling. In this paper, we use LSTM as a state space realization and compare it with linear state space modeling and statistical learning methods, including N4SID, partial least squares, the Lasso, and support vector regression. Two case studies on an industrial 660 MW boiler and a debutanizer column process indicate that LSTM underperforms all other methods. LSTM is shown to be capable of outperforming linear methods for a simulated reactor process with severely excited nonlinearity in the data. By dissecting the sub-components of a simple LSTM model, the effectiveness of the LSTM gates and nonlinear activation functions is scrutinized. • LSTM is implemented as a nonlinear Kalman filter for dynamic inferential modeling. • LSTM is benchmarked with N4SID, PLS, Lasso, and SVR statistical learning methods. • Results on a boiler emission and debutanizer and a CSTR process are reported. • The results show that LSTM underperform N4SID and statistical learning methods. • With severe nonlinearity in the simulated CSTR data, LSTM underperforms SVR. [ABSTRACT FROM AUTHOR]

After introducing process data analytics using latent variable methods and machine learning, this paper briefly review the essence and objectives of latent variable methods to distill desirable components from a set of measured variables. These latent variable methods are then extended to modeling high dimensional time series data to extract the most dynamic latent time series, of which the current values are best predicted from the past values of the extracted latent variables. We show with an industrial case study how real process data are efficiently and effectively modeled using these dynamic methods. The extracted features reveal hidden information in the data that is valuable for understanding process variability. [ABSTRACT FROM AUTHOR]

• A statistical learning procedure for Dow data challenge problem (Braun et al., 2020) is presented that integrates process knowledge in all steps from pre-processing and model interpretation. • An accurate inferential sensor model is built with online bias learning based on new data to predict the impurity in the product stream with apparent drifts. • Least angle regression solution (LARS) is shown to select only one variable among a set of collinear variables. • We report the detection of an equipment-switching operation in the data and interpolations found in the impurity data, which leads to unique data pre-processing measures. • Using a softplus function, we propose a method to deal with non-negative physical property modeling. In this paper, we propose a statistical learning procedure that integrates process knowledge for the Dow data challenge problem presented in Braun et al. (2020). The task is to build an accurate inferential sensor model to predict the impurity in the product stream with apparent drifts. The proposed method consists of i) process data exploratory analysis, ii) a method for variable selection, iii) a method to deal with non-negative physical property modeling using a softplus function; and iv) a method for online bias updating based on known data. We make use of process operation knowledge in all steps of data analytics, including exploratory analysis and feature selection. We report the detection of equipment-switching operations in the data and interpolations found in the impurity data. Partial least squares (PLS) and least angle regression solution (LARS) are adopted to model the data with strong collinearity. Pros and cons of LARS and PLS are given with practical implications. [ABSTRACT FROM AUTHOR]

• A novel procedure for diagnosing and troubleshooting plant-wide anomalies is proposed using dynamic embedded latent feature analysis. • Composite loadings are proposed for root cause analysis of an interested feature that improves the traditional bi-plots of loadings. • A dynamic inner canonical analysis algorithm with exogenous variables is proposed, which removes the impact of uninterested features. • A thorough application to an industrial plant with datasets before and after maintenance is presented to illustrate the proposed troubleshooting method. Plant-wide process data are usually high dimensional with dynamics residing in a reduced dimensional latent space. In this paper, we propose a novel procedure for diagnosing and troubleshooting plant-wide process anomalies using dynamic embedded latent feature analysis (DELFA). To remove the impact of external disturbances or exogenous variables, a dynamic inner canonical correlation analysis algorithm with exogenous variables is proposed. Composite loadings and composite weights are derived and applied for diagnosing a feature that is contained in several latent variables. The dynamic embedded latent features are usually related to poor control performance or malfunctioning control instrumentation. The proposed DELFA procedure with dynamic latent scores and composite loadings is applied to two industrial datasets of a chemical plant before and after a troubled control valve was fixed. The case study demonstrates convincingly that latent dynamic features are powerful for troubleshooting of process anomalies and diagnosing their causes in a plant-wide setting. [ABSTRACT FROM AUTHOR]

Abstract: This paper provides an overview of the state of the art of subspace identification methods for both open-loop and closed-loop systems. Practical considerations and future directions are given at the end of the paper. [Copyright &y& Elsevier]

Canonical correlation analysis (CCA) and projection to latent structures (PLS) are popular statistical approaches for process modeling and monitoring. CCA focuses on the correlation structure only, while PLS focuses on maximizing the covariance between process variables X and quality variables Y. In this article, a dynamic regularized latent variable regression (DrLVR) algorithm is proposed for dynamic data modeling and monitoring. DrLVR aims to maximize the projection of quality variables on the dynamic latent spaces of the process variables. A regularization term is incorporated into DrLVR to handle the collinearity issues. The dynamic monitoring scheme based on the DrLVR model is also developed. Both numerical simulations and the Tennessee Eastman process data are employed to demonstrate the effectiveness of DrLVR. [ABSTRACT FROM AUTHOR]

Abstract: This paper presents an overview of results and future challenges related to temperature control and cost optimization in building energy systems. Control and economic optimization issues are discussed and illustrated through simulation examples. The paper concludes with results from model predictive control solutions and identification of important directions for future work. [Copyright &y& Elsevier]

Chemical process plant safety, production specifications, environmental regulations, operational constraints, and plant economics are some of the main reasons driving an upward interest in research and development of more robust methods for process monitoring and control. Principal component analysis (PCA) has long been used in fault detection by extracting relevant information from multivariate chemical data. The recent success of wavelets and multi-scale methods in chemical process monitoring and control has catalyzed an interest in the investigation of wavelets based methods for fault detection. In the present work, multi-scale principal component analysis (MSPCA) is used for fault detection and diagnosis. MSPCA simultaneously extracts both, cross correlation across the sensors (PCA approach) and auto-correlation within a sensor (wavelet approach). Using wavelets, the individual sensor signals are decomposed into approximations and details at different scales. Contributions from each scale are collected in separate matrices, and a PCA model is then constructed to extract correlation at each scale. The multi-scale nature of MSPCA formulation makes it suitable to work with process data that are typically non-stationary and represent the cumulative effect of many underlying process phenomena, each operating at a different scale. The proposed MSPCA approach is able to outperform the conventional PCA based approach in detecting and identifying real process faults in an industrial process, and yields minimum false alarms. Additionally, the advantage of MSPCA, over the traditional PCA approach for sensor validation, is also demonstrated on an industrial boiler data set. [Copyright &y& Elsevier]

Abstract: Nonlinear state prediction is of crucial importance to design controllers for nonlinear processes with input time delay. In this paper, the extended nonlinear state predictor (ENSP) we proposed is first outlined, which is used to predict the future states of a class of nonlinear processes with input time delay. A new concept of strong tracking predictor (STP) is then proposed, and an orthogonality principle is given as a criterion to design the STP. On the basis of the orthogonality principle, the ENSP is modified, which results in a STP. After the detailed STP algorithm is presented, we prove that the STP is locally asymptotically convergent for a class of nonlinear deterministic processes if some sufficient conditions are satisfied. In the presence of measurement noise, it is further proved that the proposed STP is exponentially bounded under certain conditions. Finally, computer simulations with a MIMO nonlinear model are presented, which illustrate that the proposed STP can predict accurately the future states of a class of nonlinear time delay processes no matter whether the states change suddenly or slowly, in addition, it has definite robustness against model/plant mismatches. [Copyright &y& Elsevier]