Search

Due to its incorporation of prior process knowledge and handling of uncertainty in a probabilistic sense, Bayesian network (BN)‐based soft‐sensors have shown significant advantages compared to conventional regression‐based soft‐sensors. In the literature, these soft‐sensors are developed under the assumption that the process is operated around certain operating points. Due to the time‐varying nature of the process, the prediction performance of the existing BN‐based soft‐sensors may deteriorate over time. To account for this, in the current work, adaptive Bayesian inference using a random walk model for both process and sensor drift is proposed. The efficacy of the proposed approaches is demonstrated through simulation and industrial case studies, and they are shown to perform better than the conventional bias updated approach. [ABSTRACT FROM AUTHOR]

Data-driven causal modeling approaches find application in process data analysis and control. Process systems can be time-varying in nature; however, the existing literature for data-driven causal modeling hardly accounts for this. In this article, we present a causal modeling approach for the time-varying systems. It relies on the time-varying parameter models (TVPMs), estimated under the variational Bayesian expectation-maximization (VBEM) framework. We incorporate a hypothesis switching procedure followed by the estimation that allows us to infer the time-varying strengths of the causal influence of the inputs on the outputs of the system. We illustrate the proposed approach using the production data from steam-assisted gravity drainage (SAGD) wells. The proposed approach was found to extract consistent causal models across multiple cases of the studied system and outperform the time-invariant models and TVPMs estimated under the maximum likelihood framework. [ABSTRACT FROM AUTHOR]

Conventional methods for distributed monitoring commonly assume that complete process measurements are available. However, the problem of missing data is often encountered in the monitoring of large-scale multiunit processes. This paper proposes an approach based on a neighborhood variational Bayesian principal component analysis (NVBPCA) and canonical correlation analysis (CCA) for the efficient distributed monitoring of multiunit processes in the presence of missing data. Missing observations for a local unit are reconstructed through NVBPCA by considering information from both local and neighboring units. A CCA-based local monitor, which identifies the status of the local unit and the type of a detected fault using information from both the local and neighboring units, is then developed. The NVBPCA–CCA approach has a better performance since its missing data handling and local monitor construction consider information from both the local and neighboring units. The efficiency of the proposed monitoring method is demonstrated through its application in a numerical example and an industrial tail gas treatment process. [ABSTRACT FROM AUTHOR]

Measurements quality is important for process systems engineering. In this paper, an estimation scheme is proposed in the state-space form to monitor the degree of accuracy of measurements within a predefined horizon. Under the assumption that all the sensors are uncorrelated with each other, the distribution of measurement noise covariance as well as the distribution of state vector are estimated simultaneously. The key technique is to approximate the true posterior distribution by two independent proposal distributions using the variational Bayesian inference. It is shown that the proposed algorithm provides not only a complete picture of the working status of each sensor, but also satisfied estimates of the hidden states in the presence of faulty signals. Numerical examples with a moving target tracking model and a quadrate water tank experiment are conducted to demonstrate that the proposed method exhibits better performance than the existing methods, and even a small fluctuation of sensors can be accurately captured by the proposed algorithm. [ABSTRACT FROM AUTHOR]

This paper develops a variational Bayesian approach for identifying ARX models with missing observations and varying time-delays. The outputs of the ARX models are subject to both slow sampling rates and communication delays. The unknown missing observations which are used in the variational Bayesian approach can be estimated by a modified Kalman filter, and based on the estimated missing observations and available data, the unknown parameters and the varying time-delays can be estimated by using the variational Bayesian approach. The simulation results demonstrate that the variational Bayesian method is effective. [ABSTRACT FROM AUTHOR]

In recent decades, soft sensors have been profoundly studied and successfully applied to predict critical process variables in real‐time. While dealing with various application scenarios, data‐driven methods with representation learning possess great potentials. Latent features are formulated in these approaches to predict outputs from correlated input variables. In this study, a probabilistic framework of feature extraction is proposed in the context of process data analysis. To address switching behaviors in industrial processes, multiple emission models are utilized to construct latent space. To address temporal correlations from continuously operating processes, a dynamic model is implemented in latent space. Bayesian learning strategy is then developed for parameters estimation, where modeling preferences and uncertainties from multiple models are considered. The effectiveness and practicability of the proposed feature extraction algorithm are illustrated through numerical simulations, as well as an industrial case study. © 2018 American Institute of Chemical Engineers AIChE J, 64: 2037–2051, 2018 [ABSTRACT FROM AUTHOR]

The finite impulse response (FIR) filter and infinite impulse response filter including the Kalman filter (KF) are generally considered as two different types of state estimation methods. In this paper, the sequential Bayesian philosophy is extended to a filter design using a fixed amount of most recent measurements, with the aim of exploiting the FIR structure and unifying some basic FIR filters with the KF. Specifically, the conditional mean and covariance of the posterior probability density functions are first derived to show the FIR counterpart of the KF. To remove the dependence on initial states, the corresponding likelihood is further maximized and realized iteratively. It shows that the maximum likelihood modification unifies the existing unbiased FIR filters by tuning a weighting matrix. Moreover, it converges to the Kalman estimate with the increase of horizon length, and can thus be considered as a link between the FIR filtering and the KF. Several important properties including stability and robustness against errors of noise statistics are illustrated. Finally, a moving target tracking example and an experiment with a three degrees-of-freedom helicopter system are introduced to demonstrate effectiveness. [ABSTRACT FROM AUTHOR]

In this article, two layer mixture Bayesian probabilistic principal component analyser model is developed and proposed for fault detection. It is suitable for the data driven process monitoring applications where data with non-Gaussian distribution and temporal correlations are encountered. Model development involves modifying the original observation matrix to make it suitable for building dynamic models and followed by two stages of estimation. In the first stage, the data is divided into a manageable number of clusters and in the second stage, a mixture model is built over each cluster. This strategy provides a scalable mixture model that can have multiple local models. It has the potential to provide a parsimonious model and be less susceptible to local optima compared to the existing approaches that build mixture models in a single stage. Dimension reduction during the estimation is automated using the Bayesian regularization approach. The proposed model essentially provides a probability density function for the training data. It is deployed for fault detection and the performance highlights are demonstrated in two real datasets, one is from the oil sands industry and the other is a publicly available experimental dataset. [ABSTRACT FROM AUTHOR]

Data-driven techniques such as principal component analysis (PCA) have been widely used to derive predictive models from historical data and applied for quality prediction in industry. Motivated by reducing data collinearity and extracting informative driving forces behind data, latent variable models are explored to facilitate the prediction by regressing data on a set of extracted features. In this paper, a novel learning strategy is proposed to build dynamic features under a full Bayesian framework, incorporating data information and prior knowledge of process dynamics. Unlike the traditional PCA that extracts features based on variances explained, in this paper, the latent features are extracted with the guidance of preferred velocities of nominal variations. By applying Bayesian learning algorithms, parameters are estimated with probability distributions accounting for corresponding uncertainties, and the number of latent features can be automatically determined by the variational Bayesian inference algorithm. The effectiveness and practicability of this Bayesian dynamic feature regression are demonstrated through simulated examples as well as an industrial case study. [ABSTRACT FROM PUBLISHER]

Detection and quantification of interactions between the loops of a multivariable system are of interest for different purposes, such as control system design, optimization, fault diagnosis, and performance assessment. This paper proposes a new method for interaction analysis based on decomposing the estimated transfer function between variables in the form of impulse response coefficients. The method not only provides an estimation of the direct (feedback and interaction free) transfer function between the variables, but also provides a measure of strength of all the indirect paths connecting variables together individually. The advantage of the method is that it provides a complete picture of the different paths through which variables can influence each other along with an estimation of the energy transferred through each path independently. The analysis is performed by estimating structural vector autoregressive models under Bayesian framework. Bayesian approach provides certain advantages in terms of dealing with high dimensional variables and overparameterization problem. An appropriate design of the prior probability for the model parameters also better ensures convergence to a physically interpretable model. A procedure to design the prior distribution for the model parameters is presented in this paper. [ABSTRACT FROM AUTHOR]

Just-in-time learning (JITL) is one of the most widely used strategies for soft sensor modeling in nonlinear processes. However, traditional JITL methods have difficulty in dealing with data samples that contain missing values. Meanwhile, data noises and uncertainties have not been taken into consideration for relevant sample selection in existing JITL approaches. To overcome these problems, a new probabilistic JITL (P-JITL) framework is proposed in this brief. In P-JITL, variational Bayesian principal component analysis is first utilized to handle missing values and extract Gaussian posterior distributions of latent variables. Then, symmetric Kullback–Leibler divergence is creatively employed to measure the dissimilarity of two distributions for relevant sample selection in the JITL framework. Finally, a nonlinear regression model, Gaussian process regression, is carried out to model the nonlinear relationship between the output and the extracted latent variables. In this way, the proposed probabilistic JITL (P-JITL) is able to deal with missing data and select relevant samples more accurately. To evaluate the effectiveness and flexibility of P-JITL, comparative studies between P-JITL and traditional deterministic JITL (D-JITL) are carried out on a numerical example and an industrial application example, in which missing data are simulated with percentages from 0% to 50%. The results show that P-JITL can provide more accurate prediction accuracy than D-JITL in each scenario considered. [ABSTRACT FROM PUBLISHER]

Large-scale plant-wide processes have become more common and monitoring of such processes is imperative. This work focuses on establishing a distributed monitoring scheme incorporating multivariate statistical analysis and Bayesian method for large-scale plant-wide processes. First, the necessity of distributed monitoring is demonstrated by theoretical analysis on the impact of process decomposition on multivariate statistical process monitoring performance. Second, a stochastic optimization algorithm-based performance-driven process decomposition method is proposed which aims to achieve the best possible monitoring performance from process decomposition aspect. Based on the obtained sub-blocks, local monitors are established to characterize local process behaviors, and then a Bayesian fault diagnosis system is established to identify the underlying process status of the entire process. The proposed distributed monitoring scheme is applied on a numerical example and the Tennessee Eastman benchmark process. Comparison results to some state-of-the-art methods indicate the efficiency and feasibility. [ABSTRACT FROM AUTHOR]

Conventional Bayesian fault diagnosis assumes that all measurements are available synchronously; however, this condition does not always hold in practical industry because a process can be characterized by multiple sampling or transmitting rates. This paper introduces a Bayesian fault diagnosis system incorporating both historical and online information to address the asynchronous measurement problem. First, the Expectation Maximization approach is utilized to deal with the historical asynchronous measurements; second, the online incomplete measurements are handled through a Bayesian marginalization method within a moving horizon. Then, a Bayesian diagnosis system revealing both the underlying fault status of the whole plant and the unavailable statuses of the corresponding local units is established, which is more robust for practical application. The proposed scheme is tested on a numerical example, the distributed monitoring problem of Tennessee Eastman benchmark process, and the distributed monitoring problem of an industrial tail gas treatment plant. Monitoring results demonstrate the feasibility and efficiency of the proposed approach. [ABSTRACT FROM PUBLISHER]

Principal component analysis (PCA) serves as the most fundamental technique in multivariate statistical process monitoring. However, other than determining contributions to a fault from each variable based on the pre-selected major principal components (PCs), the PCA-based fault diagnosis with an optimal selection of PCs is seldom investigated. This paper presents a novel Gaussian mixture model (GMM) and optimal principal components (OPCs)-based Bayesian method for efficient multimode fault diagnosis. First, the GMM and Bayesian inference is utilized to identify the operating mode, and then local PCA model is established in each mode. Second, given that the various principal components (PCs) may contain distinct fault signatures, the behavior of each PC in local PCA is examined and the OPCs are selected through stochastic optimization algorithm. Based on the OPCs, a Bayesian diagnosis system is then formulated to identify the fault statuses in a probability manner. Performance of GMM–OPC Bayesian diagnosis is examined through a numerical example and the Tennessee Eastman challenge process. The efficiency and feasibility are demonstrated. [ABSTRACT FROM AUTHOR]

Multivariate statistical process monitoring involves dimension reduction and latent feature extraction in large-scale processes and typically incorporates all measured variables. However, involving variables without beneficial information may degrade monitoring performance. This study analyzes the effect of variable selection on principal component analysis (PCA) monitoring performance. Then, it proposes a fault-relevant variable selection and Bayesian inference-based distributed method for efficient fault detection and isolation. First, the optimal subset of variables is identified for each fault using an optimization algorithm. Second, a sub-PCA model is established in each subset. Finally, the monitoring results of all of the subsets are combined through Bayesian inference. The proposed method reduces redundancy and complexity, explores numerous local behaviors, and provides accurate description of faults, thus improving monitoring performance significantly. Case studies on a numerical example, the Tennessee Eastman benchmark process, and an industrial-scale plant demonstrate the efficiency. [ABSTRACT FROM AUTHOR]

Process measurements collected from daily industrial plant operations are essential for process monitoring, control, and optimization. However, those measurements are generally corrupted by errors, which include gross errors and random errors. Conventionally, those two types of errors were addressed separately by gross error detection and data reconciliation. Solving the simultaneous gross error detection and data reconciliation problem using the hierarchical Bayesian inference technique is focused. The proposed approach solves the following problems in a unified framework. First, it detects which measurements contain gross errors. Second, the magnitudes of the gross errors are estimated. Third, the covariance matrix of the random errors is estimated. Finally, data reconciliation is performed using the maximum a posteriori estimation. The proposed algorithm is applicable to both linear and nonlinear systems. For nonlinear case, the algorithm does not involve any linearization or approximation steps. Numerical case studies are provided to demonstrate the effectiveness of the proposed method. © 2015 American Institute of Chemical Engineers AIChE J, 61: 3232-3248, 2015 [ABSTRACT FROM AUTHOR]

Principal component analysis has been widely used in the process industries for the purpose of monitoring abnormal behaviour. The process of reducing dimension is obtained through PCA, while T -tests are used to test for abnormality. Some of the main contributions to the success of PCA is its ability to not only detect problems, but to also give some indication as to where these problems are located. However, PCA and the T -test make use of Gaussian assumptions which may not be suitable in process fault detection. A previous modification of this method is the use of independent component analysis (ICA) for dimension reduction combined with kernel density estimation for detecting abnormality; like PCA, this method points out location of the problems based on linear data-driven methods, but without the Gaussian assumptions. Both ICA and PCA, however, suffer from challenges in interpreting results, which can make it difficult to quickly act once a fault has been detected online. This paper proposes the use of Bayesian networks for dimension reduction which allows the use of process knowledge enabling more intelligent dimension reduction and easier interpretation of results. The dimension reduction technique is combined with multivariate kernel density estimation, making this technique effective for non-linear relationships with non-Gaussian variables. The performance of PCA, ICA and Bayesian networks are compared on data from an industrial scale plant. [ABSTRACT FROM AUTHOR]

This article develops a Bayesian method for fault detection and isolation using a sparse reconstruction framework. The normal/training data is assumed to follow a signal-plus-noise model, and an indicator matrix is used to show whether the test data is from a faulty process. The distribution of the indicator matrix is modeled by a Laplacian distribution, which forces the indicator matrix to be a sparse one, and a Gibbs sampler is derived to obtain the estimation/reconstruction of the indicator matrix, the unobserved signals, and other parameters like signal mean, covariance, and noise variance. The faulty variables can then be detected and isolated by inspecting whether corresponding rows of the indicator matrix are zero. The proposed Bayesian approach is data driven; it allows for simultaneous fault detection and isolation. A simulation study and an industrial case study are used to test the performance of the proposed method. Copyright © 2015 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Many performance monitoring algorithms (or monitors) have been developed to assess control performance and detect problems with specific components; however, these algorithms monitor single components as stand-alone experts and can be influenced by other problems that they were not meant to detect. Thus, the occurrence of a problem can lead to flood of abnormal monitor outputs and alarms which can be difficult to interpret. This work focuses on how to combine information from the many different monitoring algorithms and some of process knowledge in order to obtain a more reliable diagnosis. While traditional statistical or data-based methods need data from all abnormal cases that they should diagnose/isolate, this work focuses on how to improve the Bayesian control loop diagnosis by integrating process knowledge and training data when some of the abnormality data are sparse or not available in historical database. Simulation of the proposed Bayesian diagnostic system on the Tennessee Eastman challenge problem is presented. It is demonstrated that the diagnosis is possible even when there are no training data (or only few samples) from some abnormalities. [ABSTRACT FROM PUBLISHER]

Most existing methods for the stateestimation in batch processesare similar to those for continuous processes, and these methods usuallyonly consider the state dynamics within a single batch and ignorethe dynamics across batches. In this paper, the state estimation inbatch processes is investigated based on a two-dimensional state-spacemodel by employing the Bayesian recursive algorithm. In addition tothe dynamics along the time dimension (the dynamics within a singlebatch), the batch process is also characterized by the dynamics alongthe batch dimension (batch-to-batch dynamics). In the proposed method,both the batch-to-batch dynamics and the dynamics within a singlebatch are taken into account. The current state is dependent on theprevious states both along the time dimension and along the batchdimension, so the filtering and smoothing for previous batches shouldbe performed before doing current state estimation. In this way, theinformation on measurements from the previous batches as well as fromthe current batch can be incorporated into the estimation. The proposedmethod is illustrated and evaluated through a simple numerical exampleas well as a simulated two-state batch reaction process. [ABSTRACT FROM AUTHOR]

Highlights: [•] Analytical Bayesian solution to change point detection problem of multivariate data with both single and multiple changes. [•] Expectation and Maximization (EM) methods for change point detection problem of multivariate data with both single and multiple changes. [•] Comparative study between Bayesian approach and EM approach. [ABSTRACT FROM AUTHOR]

Abstract: In many industrial plants, development and implementation of advanced monitoring and control techniques require real-time measurement of process quality variables. However, on-line acquisition of such data may involve difficulties due to inadequacy of measurement techniques or low reliability of measuring devices. To overcome the shortcomings of traditional instrumentation, inferential sensors have been designed to infer process quality indicators from real-time measurable process variables. In recent years, due to the demonstrated advantages of Bayesian methods, interest in investigating the application of these methods for design of inferential sensors has grown. However, the potential of Bayesian methods for inferential modeling practices in the process industry has not yet been fully realized. This paper provides a general introduction to the main steps involved in development and implementation of industrial inferential sensors, and presents an overview of the relevant Bayesian methods for inferential modeling. [Copyright &y& Elsevier]

Abstract: Formulated under the state space framework, most previous methods for the state estimation typically treated batch processes in the same way as continuous ones, and only considered the state transition within a single batch. Considering that the initial state of the current batch is often related to that of the previous one, this paper incorporates the information of the previous batches into the estimation of the current state, where the filtering and smoothing for the previous batch is implemented and then the initial state of the current batch is estimated by treating the smoothed initial state of the previous batch as a “measurement”. To deal with the nonlinear and non-Gaussian property of batch processes, the particle filter method is employed as the key algorithm for filtering and smoothing. In addition, in order to make full use of various measurements, the case of missing data is considered during the implementation of the particle filter algorithm. The proposed method is illustrated and evaluated through the simulation on a penicillin fed-batch fermentation process. [Copyright &y& Elsevier]

Abstract: Control-loop performance assessment methods have been evolving over the past two decades, with many different monitor algorithms being used to single out specific problems and determine the operating mode. However, a change in operating mode may affect multiple monitors, resulting in the possibility of conflicting assessments. Data-driven Bayesian methods were previously proposed which use multiple monitors to yield probabilistic assessments; however, training data for Bayesian methods requires complete knowledge of underlying operational modes. This paper proposes an approach based on proportionality parameters θ to address the problem of incomplete mode information in the training data; values in θ can be used to fill in missing information, and by varying θ one can determine the boundaries on a probabilistic diagnosis. Two diagnostic approaches are considered: the first type is direct probability approach, which can only be applied when historical data on the operation mode is sufficient and representative. The second type is the likelihood approach which can be applied to more general cases, including when the historical data is too limited to adequately represent mode frequency. In order to represent mode frequency, the likelihood approach takes into account prior probabilities of operating modes. The proposed methods are evaluated in two simulated chemical processes. [Copyright &y& Elsevier]

In the context of process industries, outlying observations mostly represent a large random error resulting from irregular process disturbances, instrument failures, or transmission problems. Statistical analysis of process data contaminated with outliers may lead to biased parameter estimation and plant-model mismatch. The problem of process identification in the presence of outliers has received great attention and a wide variety of outlier identification approaches have been proposed. However, there is a great need to seek for more general solutions and a robust framework to deal with different types of outliers. The main objective of this work is to formulate and solve the robust process identification problem under a Bayesian framework. The proposed solution strategy not only yields maximum a posteriori estimates of model parameters but also provides hyperparameters that determine data quality as well as prior distribution of model parameters. Identification of a simulated continuous fermentation reactor is considered to show the effectiveness and robustness of the proposed Bayesian framework. The advantages of the method are further illustrated through an experimental case study of a pilot-scale continuous stirred tank heater. © 2012 American Institute of Chemical Engineers AIChE J, 59: 845-859, 2013 [ABSTRACT FROM AUTHOR]

Abstract: In the context of process industries, online monitoring of quality variables is often restricted by inadequacy of measurement techniques or low reliability of measuring devices. Therefore, there has been a growing interest in the development of inferential sensors to provide frequent online estimates of key process variables on the basis of their correlation with real-time process measurements. Representation of multi-modal processes is one of the challenging issues that may arise in the design of inferential sensors. In this paper, Bayesian procedures for the development and implementation of adaptive multi-model inferential sensors are presented. It is shown that the application of a Bayesian scheme allows for accommodating the overlapping operating modes and facilitating the inclusion of prior knowledge. The effectiveness of the proposed procedures are first demonstrated through a simulation case study. The efficacy of the method is further highlighted by a successful industrial application of an adaptive multi-model inferential sensor designed for real-time monitoring of a key quality variable in an oil sands processing unit. [Copyright &y& Elsevier]

This study presents the design of soft sensors for estimation of bitumen froth quality in an oil sands natural froth lubricated (NFL) transportation process. One of the most important quality indexes for bitumen froth is the water content. Due to the variation in oil sands composition and the nature of multi-phase process conditions, existing hardware sensors are not reliable enough to provide on-line accurate water content measurement. Laboratory analysis result is obtained off-line with large sampling interval and irregular time delay. Therefore, it is not sufficient for real-time monitoring and control. To overcome these limitations, Bayesian information synthesis approach is proposed to fuse all the existing information to produce more reliable and more accurate real-time froth quality information. This technique has been applied in Syncrude Canada Extraction operations; both monitoring and control performance illustrate the promising perspectives of the proposed approach. © 2012 Canadian Society for Chemical Engineering [ABSTRACT FROM AUTHOR]

Water temperature is an important factor affecting aquatic life within the stream environment. Cold water species, such as salmonids, are particularly susceptible to elevated water temperatures. This paper examines the economic value of short-term water temperature forecasts for salmonid management. Forecasts may have economic value if they allow the water resource manager to make better water allocation decisions. This study considers two applications: water releases for management of Chinook salmon in the Klamath River and leasing water from agriculture for management of steelhead trout in the John Day River. We incorporate biophysical models and water temperature distribution data into a Bayesian framework to simulate changes in fish populations and the corresponding benefit from recreational fishing and opportunity cost of water under different temperature forecast accuracies. Simulation results indicate that use of the forecasts results in increased fish production and that marginal costs decline and net benefits increase as forecast accuracy increases, suggesting that provision and use of such stream temperature forecasts would have value to society. ( JEL Q22, Q25, Q28, Q50) [ABSTRACT FROM AUTHOR]

Abstract: This paper is concerned with developing an online algorithm for detecting and estimating systematic errors (gross errors) in mass and energy balances from measurement data. This method has its application in diagnosing problems in an oil sands process. Conventional techniques for detecting gross errors presently exist for offline application. The proposed online method entitled Dynamic Bayesian Gross Error Detection (DBGED) is a dynamic Bayesian analogue of traditional gross error detection, and can be considered as a type of Switching Kalman Filter. As such, related topics such as Kalman Filtering, observability and Dynamic Bayesian Inference are discussed. In addition to detecting gross errors, the DBGED also estimates detected gross error magnitudes in real time (as an augmented state variable) so that future measurements can be corrected. When the estimate converges to yield satisfactory prediction errors, gross error estimation is stopped and instruments are corrected with a constant gross error correction term. DBGED performance is demonstrated through a simulation example and an example of an industrial application. [Copyright &y& Elsevier]

Abstract: Conventional Bayesian methods commonly assume that the evidences are temporally independent. This condition, however, does not hold for most engineering problems. With the evidence transition information being considered, the temporal information can be synthesized within the Bayesian framework to improve diagnosis performance. In this paper, the important evidence dependency problem is solved by a data-driven Bayesian approach with consideration of evidence transition probability. The applications in a simulated distillation column and a pilot scale process are presented to demonstrate the data dependency handling ability of the proposed control loop performance diagnosis approach. [Copyright &y& Elsevier]

The article presents a proposed data-driven Bayesian approach to determine problematic control performance and verify its effectivity through simulations and industrial application example. The approach uses the concept of missing pattern as it analyzes information from different monitoring algorithms and uses the underlying complete evidence matrix (UCEM) to solve the missing data problems. Simulated binary distillation column and an industry process verified the proposed method.

Abstract: There exist many algorithms for control performance monitoring. There are also many algorithms available for process monitoring. There are, however, few methods available for synthesis of various monitoring techniques to form a diagnosing system for optimal decision making. This paper is concerned with establishing and demonstrating a novel probabilistic diagnostic framework for control loop monitoring. The new framework possesses a number of desired properties, including, for example, quantitative probabilistic diagnosis, flexibility in synthesizing different monitoring techniques, robustness in the presence of missing data or missing variables, ease of expanding or shrinking the diagnosing system, ability to incorporate a priori process knowledge, and capability for decision making. As the backbone of the proposed framework, the emerging Bayesian methods are introduced and shown to be the appropriate tools for solving the problem of concern. Several representative control loop diagnostic problems are formulated under the Bayesian framework and their solutions are demonstrated through examples. The experiences and challenges learned from industrial applications of Bayesian methods are summarized and some of future research directions are discussed. [Copyright &y& Elsevier]

Abstract: Building mathematical models is a common task in process systems engineering, which requires estimation of model parameters as the final step of modeling exercise. Model based experimental design has evolved as a potential statistical tool for reducing uncertainties in parameter estimates. Often a huge volume of process information is generated as an end result of an experimental design. Designing optimal experiments based on current or prior process knowledge is still an open research problem. This paper deals with how information, available a priori, can be organized and systematically used for designing robust Bayesian dynamic experiments, in the presence of process constraints. The designed experiments are ‘robust’ to a poor choice of nominal parameter values. Several novel techniques for organizing a priori process knowledge are explored from a theoretical view point. The influence of proposed prior designs on parameter estimates is demonstrated on a semi-continuous baker''s yeast fermenter problem. [Copyright &y& Elsevier]

Abstract: This paper deals with the issues associated with the development of data-driven models as well as model update strategy for soft sensor applications. A practical yet effective solution is proposed. Key process variables that are difficult to measure are commonly encountered in practice due to limitations of measurement techniques. Even with appropriate instruments, some measurements are only available through off-line laboratory analysis with typical sampling intervals of several hours. Soft sensors are inferential models that can provide continuous on-line prediction of hidden variables; such models are capable of combining real-time measurements with off-line lab data. Due to the prevalence of plant-model mismatch, it is important to update the model using the latest reference data. In this paper, parameters of data-driven models are estimated using particle filters under the framework of expectation–maximization (EM) algorithms. A Bayesian methodology for model calibration strategy is formulated. The proposed framework for soft sensor development is applied to an industrial process to provide on-line prediction of a quality variable. [Copyright &y& Elsevier]