Tennessee Eastman Process

In order to achieve an optimum and successful operation of an industrial process, it is important firstly to detect upsets, equipment malfunctions or other abnormal events as early as possible and secondly to identify and remove the cause of those events. Univariate and multivariate statistical process control methods have been widely applied in process industries for early fault detection and localization. The primary objective of the proposed research is the design of an anomaly detection and visualization tool that is able to present to the shift operator -and to the various levels of plant operation and company management -an early, global, accurate and consolidated presentation of the operation of major subgroups or of the whole plant, aided by a graphical form. Piecewise Aggregate Approximation (PAA) and Symbolic Aggregate Approximation (SAX) are considered as two of the most popular representations for time series data mining, including clustering, classification, pattern discovery and visualization in time series datasets. However SAX is preferred since it is able to transform a time series into a set of discrete symbols, e.g. into alphabet letters, being thus far more appropriate for a graphical representation of the corresponding information, especially for the shift operator. The methods are applied on individual time records of each process variable, as well as on entire groups of time records of process variables in combination with Hidden Markov Models. In this way, the proposed visualization tool is not only associated with a process defect, but it allows also identifying which specific abnormal situation occurred and if this has also occurred in the past. Case studies based on the benchmark Tennessee Eastman process demonstrate the effectiveness of the proposed approach. The results indicate that the proposed visualization tool captures meaningful information hidden in the observations and shows superior monitoring performance.

Several factors, such as human error, environmental factors, and maintenance practices, contribute to the degradation of real-world industrial systems. Predicting system dynamics is challenging and necessitates high user confidence, as these factors contribute to both aleatoric uncertainty (inherent system variability) and epistemic uncertainty (due to limited information). Decision-making and risk assessment are frequently hindered by the inability of current artificial intelligence methods to generate interpretable uncertainty estimates. To address these constraints, we propose an analysis that employs Deep Gaussian Processes (DGPs), a robust framework for generating interpretable uncertainty distributions and capturing system variability. A rigorous mathematical foundation is essential to our approach, because it enables the selection of metrics that effectively capture the system's degradation aspects. In addition to predicting the remaining useful life, these metrics, when used in conjunction with DGPs, facilitate the creation of a degradation model that is both accurate and dependable. This model also contributes to the improvement of system reliability and proactive maintenance. We demonstrate our approach's practical efficacy by validating it on a real-world industrial semolina plant with four mills. INDEX TERMS Industrial system, deep Gaussian process, uncertainty quantification, degradation modeling.

Many models and methodologies in order to predict the remaining useful life (RUL) of a critical component are investigated nowadays. However, estimating remaining useful life of multi-component systems is still an under explored area, especially when there are interdependencies among the components of these systems. Practically, prognostics can be quite complicated when there is absence of prior knowledge about these interactions. To optimize the availability and reliability of the system, it is required to embed a dependency model that implements component strength and the component's health status. In this paper, a novel approach is proposed in order to emphasize the importance of interactions between complex system's components in the RUL's calculating. The effectiveness of the approach is judged by applying it to numerical studies in order to estimate system's remaining useful life.

We present a new model for reliability analysis that is able to employ condition monitoring data in order to simultaneously monitor the latent degradation level and track failure progress over time. The method presented in this paper is a bridge between Bayesian filtering and classical binary classification, both of which have been employed successfully in various application domains. The Kalman filter is used to model a discrete-time continuous-state degradation process that is hidden and for which only indirect information is available through a multi-dimensional observation process. Logistic regression is then used to connect the latent degradation state with the failure process that is itself a discrete-space stochastic process. We present a closed-form solution for the marginal log-likelihood function and provide formulas for few important reliability measures. A dynamic cost-effective maintenance policy is finally introduced that can employ sensor signals for real-time decision-making. We finally demonstrate the accuracy and usefulness of our framework via numerical experiments.

Data-based process fault diagnosis is enhanced by using better characterization of the transient stages of the faults. After monitoring Normal operating conditions (NOC) data and projecting the model onto Abnormal operating conditions (AOC), statistical indices (T 2 and Q) are used to characterize fault transient stages by assessing faults' delay and span. Hence, a new NOC PCA model is constructed using a data subset based on the transient window and projected onto the NOC and AOC transient data. Model construction is finally achieved by applying a classification technique-Artificial Neural Networks (ANN) or Support Vector Machines (SVM), which are compared-to the

Data-based process fault diagnosis is enhanced by using better characterization of the transient stages of the faults. After monitoring Normal operating conditions (NOC) data and projecting the model onto Abnormal operating conditions (AOC), statistical indices (T 2 and Q) are used to characterize fault transient stages by assessing faults' delay and span. Hence, a new NOC PCA model is constructed using a data subset based on the transient window and projected onto the NOC and AOC transient data. Model construction is finally achieved by applying a classification technique-Artificial Neural Networks (ANN) or Support Vector Machines (SVM), which are compared-to the

Early fault detection, which reduces the possibility of catastrophic damage, is possible by detecting the change of characteristic features of the signals. The aim of this article is to detect faults in complex industrial systems, like the Tennessee Eastman Challenge Process, through on-line monitoring. The faults that are concerned correspond to a change in frequency components of the signal. The proposed approach combines the filters bank technique, for extracting frequency and energy characteristic features, and the Dynamic Cumulative Sum method (DCS), which is a recursive calculation of the logarithm of the likelihood ratio between two local segments. The method is applied to detect the perturbations that disturb the Tennessee Eastman Challenge Process and may lead the process to shut down.

Technological advancements have a broad scope in terms of experimental, analytical, field cases and numerical studies in complex petroleum engineering projects. These can be pertinent to transportation system and gathering and safety in oil and gas production. The current research was aimed at examining the challenges pertinent to safety prognostic technology as well as various ways in which it can be implemented for resolving issues in complex petroleum engineering projects. For the conduct of this research, qualitative methodology was used and primary data was assessed to present critical evaluation of the stated aim. The interviews were conducted from 10 petroleum engineers working in different public and private companies in Pakistan. The snowball technique followed by thematic analysis data analysis technique was applied for the generation of primary findings. The results of the research examined that safety prognostic technologies are significant in terms of enhancing safety, reliability and reducing the possible errors in maintenance. It has further examined that in complex engineering systems, there are multiple propagation paths to different consequences some of which might differ with respect to the most single faults.

In order to achieve an optimum and successful operation of an industrial process, it is important firstly to detect upsets, equipment malfunctions or other abnormal events as early as possible and secondly to identify and remove the cause of those events. Univariate and multivariate statistical process control methods have been widely applied in process industries for early fault detection and localization. The primary objective of the proposed research is the design of an anomaly detection and visualization tool that is able to present to the shift operator-and to the various levels of plant operation and company management-an early, global, accurate and consolidated presentation of the operation of major subgroups or of the whole plant, aided by a graphical form. Piecewise Aggregate Approximation (PAA) and Symbolic Aggregate Approximation (SAX) are considered as two of the most popular representations for time series data mining, including clustering, classification, pattern discovery and visualization in time series datasets. However SAX is preferred since it is able to transform a time series into a set of discrete symbols, e.g. into alphabet letters, being thus far more appropriate for a graphical representation of the corresponding information, especially for the shift operator. The methods are applied on individual time records of each process variable, as well as on entire groups of time records of process variables in combination with Hidden Markov Models. In this way, the proposed visualization tool is not only associated with a process defect, but it allows also identifying which specific abnormal situation occurred and if this has also occurred in the past. Case studies based on the benchmark Tennessee Eastman process demonstrate the effectiveness of the proposed approach. The results indicate that the proposed visualization tool captures meaningful information hidden in the observations and shows superior monitoring performance.

The determination of the time at which an event may take place in the future is a well-studied problem in a number of science and engineering disciplines. Indeed, for more than fifty years, researchers have tried to establish adequate methods to characterize the behaviour of dynamic systems in time and implement predictive decision-making policies. Most of these efforts intend to model the evolution in time of nonlinear dynamic systems in terms of stochastic processes; while defining the occurrence of events in terms of first-passage time

This paper addresses the problem of uncertainty quantification in system-level prognostics. To this purpose, a three-step methodology, based on the inoperability input-output model, is presented. The first step concerns the estimation of the system inoperability, using a new adapted particle filtering method, while considering the interactions between its components. The second step focuses on the long-term prediction of the system inoperability in order to determine its evolution. Finally, in the third step, a method for calculating the remaining useful life of the system, based on the system configuration, is formulated. The proposed methodology is applied on data obtained from the Tennessee Eastman Process simulations to predict the shutdown due to violation of process constraints.

Prediction of the Remaining Useful Lifetime (RUL) of the system brings down the maintenance cost, downtime and also helps to take corrective measures. This results in avoiding catastrophic events. In this Paper, The RUL prediction is done based on KalmanFilter approach with dynamic curve fitting. The desired state (which is synthetic data) of the system is estimated based on a second order Kalman Filter, where Newtonian Kinematic model is used in tracking the significant feature state of the system. The dynamic curve fitting is done based on Least Square Error Sense method. The dynamically fitted curve is extrapolated until the failure threshold is reached and subsequently RUL is estimated. The algorithm thus developed is validated by a single phase full wave bridge rectifier analogous to aircraft Transformer Rectifier Unit (TRU), to obtain the real time significant feature data. The experimental results are compared with those of developed algorithms and results show Kalman filter based algorithm is ~95% accurate.

Fault detection and diagnosis is a challenging problem for plant economics and safety. In this context, a promising OPC-based modular architecture for a Fault Diagnosis System (FDS) is designed and implemented. This FDS has been validated by performing on-line real-time diagnosis on a simulated process. The modular architecture allows openly connecting a simulator or a real process via OPC. The Tennessee Eastman Process (TEP) is used as data generator and case study, so that several abnormal operation conditions can be diagnosed by the system. The proposed architecture is discussed regarding the integration of future modules for the timely adoption of appropriate corrective actions.

In this paper, we propose an approach for failure prognosis. The approach deals with a closed-loop control system in which the actuator stochastically degrades through time. The degradation of a system disturbs and affects its characteristic parameters. This is reflected by a change in one or more of them. The latter may remain partially or totally hidden given that the type of control. The aim of this work was to estimate online the duration before the system performance requirement is no longer met. This without adding sensors. The proposed approach is based on the system behavior model. The models describing the dynamics of the parameters have been assumed to be known a priori, but degradation is assumed to be unmeasurable. It was conducted in two phases: the first used the data available on this system to estimate unmeasured states and relevant parameters which are able to characterize system performance. To carry out this phase, we used an observer. In the second phase, to esti...

The degradation of photovoltaic (PV) modules remains a major concern on the control and the development of the photovoltaic field, particularly, in regions with difficult climatic conditions. The main degradation modes of the PV modules are corrosion, discoloration, glass breaks, and cracks of cells. However, corrosion and discoloration remain the predominant degradation modes that still require further investigations. In this paper, a model-based PV corrosion prognostic approach, based on an ensemble Kalman filter (EnKF), is introduced to identify the PV corrosion parameters and then estimate the remaining useful life (). Simulations have been conducted using measured data set, and results are reported to show the efficiency of the proposed approach.

We present a new model for reliability analysis that is able to employ condition monitoring data in order to simultaneously monitor the latent degradation level and track failure progress over time. The method presented in this paper is a bridge between Bayesian filtering and classical binary classification, both of which have been employed successfully in various application domains. The Kalman filter is used to model a discrete-time continuous-state degradation process that is hidden and for which only indirect information is available through a multi-dimensional observation process. Logistic regression is then used to connect the latent degradation state with the failure process that is itself a discrete-space stochastic process. We present a closed-form solution for the marginal log-likelihood function and provide formulas for few important reliability measures. A dynamic cost-effective maintenance policy is finally introduced that can employ sensor signals for real-time decision-making. We finally demonstrate the accuracy and usefulness of our framework via numerical experiments.

International audienceIn the last two decades, prediction of the remaining useful life (RUL) of critical components and more recently of systems (SRUL) received an increasing attention. Indeed, this prediction allows scheduling predictive maintenance actions in order to avoid damages caused by unexpected failures. A major part of the studies related to failure prognostics reported in the literature considers the mission profile as a static parameter when evaluating RUL/SRUL. However in practice, operating parameters can be adjusted according to the system health state for extending the SRUL while satisfying the performance requirements. Therefore, this paper addresses the issue of adjusting the mission profile of a multicompo-nent system in order to maximize its SRUL. To do this, a variant of the inoperability input-output model is used to represent the system degradation and take into account the interactions between its components. Then, the mission profile parameters (load levels...

Le pronostic est le processus de prédiction de la durée de vie résiduelle utile (RUL) des composants, sous-systèmes ou systèmes. Cependant, jusqu'à présent, le pronostic a souvent été abordé au niveau composant sans tenir compte des interactions entre les composants et l'impact de l'environnement, ce qui peut conduire à une mauvaise prédiction du temps de défaillance dans des systèmes complexes. Dans ce travail, une approche de pronostic au niveau du système est proposée. Cette approche est basée sur un nouveau cadre de modélisation : le modèle d'inopérabilité entrée-sortie (IIM), qui permet de prendre en compte les interactions entre les composants et les effets du profil de mission et peut être appliqué pour des systèmes hétérogènes. Ensuite, une nouvelle méthodologie en ligne pour l'estimation des paramètres (basée sur l'algorithme de la descente du gradient) et la prédiction du RUL au niveau système (SRUL) en utilisant les filtres particulaires (PF), a ét...

Data-based process fault diagnosis is enhanced by using better characterization of the transient stages of the faults. After monitoring Normal operating conditions (NOC) data and projecting the model onto Abnormal operating conditions (AOC), statistical indices (T 2 and Q) are used to characterize fault transient stages by assessing faults' delay and span. Hence, a new NOC PCA model is constructed using a data subset based on the transient window and projected onto the NOC and AOC transient data. Model construction is finally achieved by applying a classification technique-Artificial Neural Networks (ANN) or Support Vector Machines (SVM), which are compared-to the

This paper addresses the problem of uncertainty quantification in system-level prognostics. To this purpose, a three-step methodology, based on the inoperability input-output model, is presented. The first step concerns the estimation of the system inoperability, using a new adapted particle filtering method, while considering the interactions between its components. The second step focuses on the long-term prediction of the system inoperability in order to determine its evolution. Finally, in the third step, a method for calculating the remaining useful life of the system, based on the system configuration, is formulated. The proposed methodology is applied on data obtained from the Tennessee Eastman Process simulations to predict the shutdown due to violation of process constraints.

Considering traditional model-based prognostics approaches, a previously defined model is required to estimate the system’s health state and then propagate it to predict the system remaining useful life (SRUL). Following a Bayesian framework, the result of this prior estimation is updated by in-field measurements without changing the model parameters. Nevertheless, in the case of prognostics at system-level, solely updating prior health state, based on the pre-determined model, is no longer sufficient because numerous mutual interactions between components cause multiple uncertainties in system degradation modeling, and then can lead to inaccurate SRUL prediction. Therefore, this paper proposes a new methodology for online joint uncertainty quantification and model estimation based on particle filtering (PF) and gradient descent (GD). In detail, the inoperability input-output model (IIM) is used to characterize system degradations considering interactions between components and effe...