Recursive data-based prediction and control of batch product quality (original) (raw)
Related papers
Data-Driven Modeling and Quality Control of Variable Duration Batch Processes with Discrete Inputs
Industrial & Engineering Chemistry Research, 2017
Batch process reactors are often used for products where quality is of paramount importance. To this end, this work addresses the problem of direct, data-driven, quality control for batch processes. Specifically, previous results using subspace identification for modeling dynamic evolution and making quality predictions are extended with two key novel contributions: first, a method is proposed to account for midbatch ingredient additions in both the modeling and control stages. Second, a novel model predictive control scheme is proposed that includes batch duration as a decision variable. The efficacy of the proposed modeling and control approaches are demonstrated using a simulation study of a poly(methyl methacrylate) (PMMA) reactor. Closed loop simulation results show that the proposed controller is able to reject disturbances in feed stock and drive the number-average molecular weight, weight-average molecular weight, and conversion to their respective set-points. Specifically, mean absolute percentage errors (MAPE) in these variables are reduced from 8.66%, 7.87%, and 6.13% under traditional PI control to 1.61%, 1.90%, and 1.67%, respectively.
Trajectory Tracking of Batch Product Quality Using Latent Variable Models
IFAC Proceedings Volumes, 2014
A practical strategy for controlling batch product quality evolution by means of latent variable models and intermittent measurements is presented. The methodology is based on the identification of data-based models using multivariate statistical methods such as Partial Least Squares (PLS). PLS is able to identify models with a reduced number of latent variables, which account for most of the process variability. The data-based models are employed along with a moving window strategy in order to predict product quality throughout the batch operating time. The predictions can be utilized within a Model Predictive Control (MPC) architecture so that trajectory tracking control can be directly applied to batch product quality. A simulated example of fed-batch aerobic growth of Saccharomyces Cerevisiae is used to demonstrate the capabilities of the proposed trajectory tracking controller.
Model-based quality monitoring of batch and semi-batch processes
2000
In this paper, a model-based inferential quality monitoring approach for a class of batch systems is investigated. Given the appropriate model form, the batch quality monitoring problem can be reduced to the problem of state estimation for batch and semi-batch processes. Because feed upsets are often a major source of disturbance in this type of system, it is shown that estimating the initial conditions can lead to improved state estimates throughout the batch as well as improved monitoring and control of enduse quality in many cases. The approach taken in this paper is to reduce the eects of the initial uncertainty resulting from feed disturbances by using algorithms designed to perform on-line smoothing of the initial conditions. First, an Extended Kalman Filterbased ®xed-point smoothing algorithm is presented and compared to a popular approach to estimating the initial conditions. Subsequently, a nonlinear optimization-based approach is introduced and analyzed. A sub-optimal on-line approximation to the optimization problem is developed and shown to be directly related to the Extended Kalman Filter-based results. Finally, some practical implementation aspects are discussed, along with simulation results from an industrially relevant example application. #
Time-invariant, Databased Modeling and Control of Batch Processes
2016
Batch reactors are often used to produce high quality products because any batch that does not meet quality specifications can be easily discarded. However, for high-value products, even a few wasted batches constitute substantial economic loss. Fortunately, databases of historical data that can be exploited to improve operation are often readily available. Motivated by these considerations, this thesis addresses the problem of direct, data-based quality control for batch processes. Specifically, two novel datadriven modeling and control strategies are proposed. The first approach addresses the quality modeling problem in two steps. To begin, a partial least squares (PLS) model is developed to relate complete batch trajectories to resulting batch qualities. Next, the so called missing-data problem, encountered when using PLS models partway through a batch, is addressed using a data-driven, multiple-model dynamic modeling approach relating candidate input trajectories to future outpu...
2012
The objective of this paper is to develop a framework that integrates two important concepts: Statistical process control (SPC) and engineering process control (EPC). Most of the literature researches on integrated SPC/EPC systems are focused into continuous process mainly with Algorithmic SPC. The integrated SPC/EPC systems in batch process control have not received the same degree of attention. In particular, there is an only Run-to-Run (RTR) control methodology application, which is mostly focused in semiconductor industry. This paper is a first of its kind in integrated SPC/EPC systems that applied in batch process and it based on a data-driven quality improvement tools. The proposed SPC/EPC integration is performed continually in two successive phases: (1) Active SPC for the batch making advance, and (2) RTR control action between batches. Control limits for critical variables are developed using information from the historical reference distribution of past successful batches. EPC application is based on the development of progressive knowledge-based rules. For a validation purpose, the proposed approach is applied to data collected from an industrial batch alkyd polymerization reactor, which evolution is monitored by measuring the overflow water weight, the acidity index and the viscosity of samples withdrawn from the reactor. This industrial process is poorly automated, subject to several disturbances, and the batches have uneven lengths. The synthesis is stopped at the maximum yield allowed by the gelation point of the cold product. Through this case study application, process engineers at the company are now able to use a valuable decision making tool when the production process is affected by certain disruptions, with obvious consequences on product quality, productivity and competitiveness.
Online Performance Monitoring and Quality Prediction for Batch Processes
IFAC Proceedings Volumes, 2004
Two different quality prediction techniques are incorporated with online MSPM through PLS modeling in this study. The first technique is based on unfolding a batch data array by preserving variable direction. An MPLS model between this matrix and vector of elapsed local batch times is developed to reflect the batch progress. More data partitions become available as the batch progresses and these partitions are rearranged into a matrix to develop local MPLS models predicting quality online. The second technique uses hierarchical PLS modeling in an adaptive manner resulting in a model that can be used to predict end-of-batch quality online. Neither technique requires estimation of future portions of variable trajectories and both are suitable for online multivariate statistical process monitoring and fault diagnosis. Case studies from a simulated fed-batch penicillin fermentation illustrate the implementation of the methodology.
Uneven batch data alignment with application to the control of batch end-product quality
ISA Transactions, 2014
Batch processes are commonly characterized by uneven trajectories due to the existence of batchto-batch variations. The batch end-product quality is usually measured at the end of these uneven trajectories. It is necessary to align the time differences for both the measured trajectories and the batch end-product quality in order to implement statistical process monitoring and control schemes. Apart from synchronizing trajectories with variable lengths using an indicator variable or dynamic time warping, this paper proposes a novel approach to align uneven batch data by identifying shortwindow PCA&PLS models at first and then applying these identified models to extend shorter trajectories and predict future batch end-product quality. Furthermore, uneven batch data can also be aligned to be a specified batch length using moving window estimation. The proposed approach and its application to the control of batch end-product quality are demonstrated with a simulated example of fed-batch fermentation for penicillin production.
Chemical Engineering Research and Design, 2009
A large number of production processes for the manufacturing of specialty chemicals, pharmaceuticals, foodstuff, and materials for microelectronics are run in batch mode. Batch processes are "simple" in terms of equipment and operation design, but are often quite complicated in terms of product quality monitoring and of production scheduling and organization. In this paper an industrial case study is presented where the challenges related to the real-time estimation of the required time to manufacture a resin and to the instantaneous product quality estimation are addressed using multivariate statistical techniques. The industrial process is poorly automated, subject to several disturbances, and the batches have uneven lengths. It is shown that stage and batch lengths can be estimated in real time with an average error that is not larger than 20% of the inherent batch-to-batch variability, whereas quality estimations can be provided within the accuracy of the hardware instrumentation, but 240 times faster. The industrial benefits deriving from the use of the proposed monitoring system have been a drastic reduction of the number of samples that need to be analyzed by the lab, prompter adjustment of the processing recipe with consequent reduction of the total processing time, and improved capability to plan the production.
Integrating data-based modeling and nonlinear control tools for batch process control
AIChE Journal, 2011
This work presents a data-based multi-model approach for modeling batch systems in which multiple local linear models are identified using partial least squares (PLS) regression and then combined with an appropriate weighting function that arises from fuzzy c-means clustering. The resulting data-based model is used to generate estimates of empirical reverse-time reachability regions (RTRRs) (defined as the set of states from where the data-based model can be driven inside a desired end-point neighborhood of the batch system) using an optimization based algorithm. The empirical RTRRs are used to formulate a computationally efficient predictive controller with inherent fault-tolerant characteristics. Simulation results of a fed-batch reactor subject to noise, disturbances, and uncertain parameters demonstrate that the empirical RTRR-based MPC design consistently outperforms PI control in both a fault-free and faulty environment. *Financial support by NSERC and the McMaster Advanced Control Consortium (MACC) is gratefully acknowledged.