Latent variable MPC for trajectory tracking in batch processes

In order to meet tight product quality specifications for batch/semi-batch processes, it is vital to monitor and control product quality throughout the batch duration. The ideal strategy is to achieve end-product quality specifications through trajectory tracking control during a batch run. However, due to the lack of in-situ sensors for continuous monitoring of batch product quality, the measurements are usually implemented by laboratory assays and are inherently intermittent. Therefore, direct trajectory tracking of batch product quality is challenging in such applications. This paper proposes a practical approach to realise trajectory tracking control of batch product quality in those situations where only intermittent measurements are available. The first step of the approach consists in identifying a projection to latent structures (PLS) model to identify a relationship between readily measured process variable trajectories and intermittently measured batch product quality. Then the identified PLS-based prediction model is transformed into recursive formulation by utilising missing data imputation algorithms. Such recursive formulation allows identified PLS-based model to be readily incorporated as a predictor into standard model predictive control (MPC) framework. Case study employing simulated fed-batch fermentation process used to manufacture penicillin was employed to illustrate the principle and the e↵ectiveness of the proposed approach.

Section: Trajectory Tracking Controlmentioning

confidence: 99%

Trajectory tracking of batch product quality using intermittent measurements and moving window estimation

Lopez-Montero

Wan

Marjanović

2015

“…Further studies in to the use of multivariate statistical models to regulate batch processes have been conducted by several research groups, including [9][10][11][12][13]. Whilst this work has been demonstrated to be successful, multivariate techniques are not without their limitations.…”

Section: Introductionmentioning

confidence: 99%

“…This is in contrast to alternative formulations found in the literature which solve the QP problem in the Latent Variable (LV) space [6,7,9,24].The drawback of the LV approach is that once the optimized points are found, it is necessary to compute the real MVT by inverting the PLS model, which can cause actuation changes that are detrimental to the yield if the PLS model is not sufficiently constrained. In addition, the proposed design uses validity restrictions inside the MVT optimization to limit the solution of the QP problem to the region within which there is confidence in the predictions made by the PLS model.…”

Section: Introductionmentioning

confidence: 99%

Multivariate batch to batch optimisation of fermentation processes incorporating validity constraints

Duran-Villalobos

Lennox

Lauri

2016

This paper presents an innovative optimisation technique, which utilises an adaptive Multiway Partial Least Squares (MPLS) model to track the dynamics of a batch process from one batch to the next. Utilising this model, an optimisation algorithm solves a quadratic cost function that identifies operating conditions for the subsequent batch that should increase yield. Hard constraints are shown to be required when solving the cost function to ensure that batch conditions do not vary too greatly from one batch to the next. Furthermore, validity constraints are imposed to prevent the PLS model from extrapolating significantly when determining new operating conditions. The capabilities of the proposed technique are illustrated through its application to two benchmark fermentation simulations, where its performance is shown to compare favourably with alternative batch-to-batch optimisation techniques.

“…LV-MPC is a model-based predictive control methodology implemented in the space of the latent variables. LV-MPC can be applied to batch [5] and continuous [6] processes. The advantages of using LV-MPC include:…”

Section: Introductionmentioning

confidence: 99%

Latent variable based model predictive control: Ensuring validity of predictions

Lauri

Sanchis

Martínez

et al. 2013

This paper presents a methodology to constrain the optimization problem in LV-MPC so that validity of predictions can be ascertained. LV-MPC is a model-based predictive control methodology implemented in the space of the latent variables and is based on a linear predictor. Provided real processes are non-linear, there is model-process mismatch, and under tight control, the predictor can be used for extrapolation. Extrapolation leads to bad predictions which deteriorates control performance, hence the interest in validity of predictions. In the proposed approach first two validity indicators on predictions are defined. The novelty in the two indicators proposed is they neglect past data, and so validity of predictions is ascertained in terms of future moves which are actually the degrees of freedom in the optimization. Second, the indicators are introduced in the optimization as constraints. Provided the indicators are quadratic, recursive optimization with linearised constraints is implemented. A MIMO example shows how ensuring validity of predictions neglecting past data can improve closed-loop performance, specially under tight control outside the identification region.