Combining Kernel Partial Least-Squares Modeling and Iterative Learning Control for the Batch-to-Batch Optimization of Constrained Nonlinear Processes

Zhang, Yingwei; Fan, Yunpeng; Zhang, Pengchao

doi:10.1021/ie1004702

Cited by 29 publications

(17 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Once the prediction significantly deviates from the desired value, the method performs mid‐course correction (MCC) via optimization and the remainders of the manipulated variable trajectories are obtained from the optimal solution . Recently, integrated batch‐to‐batch control and within‐batch online control strategies have also been presented . By combining the two methods, the integrated control strategies can obtain satisfactory performances because within‐batch online control can respond to disturbances immediately and batch‐to‐batch control can compensate for model bias.…”

Section: Batch‐to‐batch Optimization Strategymentioning

confidence: 99%

“…Flores‐Cerrillo and MacGregor presented a terminal ILC based on the PLS model and the desired final product qualities were achieved by using an iterative procedure that works in the reduced space of the latent variable model . Nonlinear data‐driven model‐based ILC and integrated ILC and within‐batch control approaches have also been recently proposed to calculate the optimal control profiles . Nonetheless, the results offer a substantial modification to the existing data‐driven model‐based ILC strategies of batch processes, which are merely set‐point tracking control algorithms …”

Section: Introductionmentioning

confidence: 99%

“…[23] Nonlinear data-driven model-based ILC and integrated ILC and within-batch control approaches have also been recently proposed to calculate the optimal control profiles. [24][25][26][27] Nonetheless, the results offer a substantial modification to the existing data-driven model-based ILC strategies of batch processes, which are merely set-point tracking control algorithms. [28] A gradient modification term was used to compensate for the cost function of optimization problems in the Integrated System Optimization and Parameter Estimation (ISOPE) method.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Self‐correcting modifier‐adaptation strategy for batch‐to‐batch optimization based on batch‐wise unfolded PLS model

2016

View full text Add to dashboard Cite

The problem of optimizing a batch process under model uncertainty using a batch‐wise unfolded PLS (BW‐PLS) model‐based modifier‐adaptation (MA) strategy is described. The main idea behind the strategy is to use measurements and iteratively modify the model to compensate for the mismatch of the necessary condition of optimality (NCO) between the plant and the model‐based optimization problem. It is proven that the popular data‐driven model‐based iterative learning control (ILC) strategy is equivalent to the proposed MA strategy using only zero‐order modifier. Inspired by the effectiveness of the ILC being enhanced by rebuilding the data‐driven model, a more elaborate model updating scheme is proposed in this paper to improve the optimization performances. The heuristic rules for choosing filtering gain matrix are also presented to further accelerate the convergence rate and reduce the variation of the cost during the period of evolution. Finally, the efficacy of the proposed MA strategy is illustrated via a simulated typical batch reaction and a simulated cobalt oxalate synthesis process.

show abstract

Section: Batch‐to‐batch Optimization Strategymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Self‐correcting modifier‐adaptation strategy for batch‐to‐batch optimization based on batch‐wise unfolded PLS model

2016

View full text Add to dashboard Cite

show abstract

“…Similar to batch‐to‐batch optimization, the basic idea behind the ILC method is to update the control profile for a new batch run by using the information obtained from previous batch runs, and the outputs converge asymptotically to the desired values. To further improve the control performances, different types of data‐driven models have been used to predict the end‐product qualities as in Flores‐Cerrillo and MacGregor, Xiong and Zhang, and Zhang et al However, these data‐driven model based ILC methods are only suitable for tracking the set‐points of the end‐product qualities …”

Section: Introductionmentioning

confidence: 99%

Combining just‐in‐time modelling and batch‐wise unfolded PLS model for the derivative‐free batch‐to‐batch optimization

2017

View full text Add to dashboard Cite

In this work, a derivative‐free batch‐to‐batch optimization method is proposed. In order to conquer the difficulties in building a first principal model, a local batch‐wise unfolded PLS (BW‐PLS) model is used to accurately describe the concerned region, and the first principal model based dynamic optimization problem is transformed into a static one. The just‐in‐time (JIT) modelling method is employed to dynamically update the local BW‐PLS model upon request, and the nonlinearity and abrupt changes from one batch run to another can be effectively resolved. Then the proposed local BW‐PLS model with JIT modelling method is integrated into the trust‐region framework. Not only can the issue of plant‐model mismatch be dealt with, but also the computation of the experimental gradients can be avoided. In addition, taking the advantages of PLS regression, the Hotelling's T2 statistic is utilized as a hard constraint to ensure the reliability of the optimal solution. Extension to handle soft inequality constraints is also included in this work. Finally, the efficacy of the proposed batch‐to‐batch optimization method is illustrated via a toy example and a simulated cobalt oxalate synthesis process under different operating conditions, and satisfied optimization performances were obtained.

show abstract

“…have been used for data processing, showing unique advantages. Currently, there are applications using some kernel methods for batch process control [11], [12]. Data of batch processes generally are of features of high dimensions, strong coupling and nonlinearities.…”

Section: Introductionmentioning

confidence: 99%

Constrained batch-to-batch optimal control for batch process based on kernel principal component regression model

Huang

Zhao

2012

2012 IEEE Fifth International Conference on Advanced Computational Intelligence (ICACI)

View full text Add to dashboard Cite

A batch-to-batch optimal control method is presented in the paper for batch process control under input constraints. Generally it is very difficult to acquire an accurate mechanistic model for a batch process. Kernel principal component regression (KPCR) technique is a nonlinear modeling method that has a better ability to deal with nonlinear data. A KPCR model based batch-to-batch optimal control strategy is developed for end-point quality control of batch process. On the basis of the linearized KPCR model, the control input is obtained by minimising a quadratic objective function concerning the end-point product quality. To ensure the safe, smooth operations of batch process, certain input constraints are taken into considered. Furthermore, the KPCR model is updated from batch-to-batch to overcome the process variations or disturbances. Numerical simulation shows that the method can improve the end-point product qualities from batch to batch under input constraints. Based on updated KPCR model, the approach has better adaptability for process variations or disturbances than the policy based on updated PCR model has.

show abstract

Combining Kernel Partial Least-Squares Modeling and Iterative Learning Control for the Batch-to-Batch Optimization of Constrained Nonlinear Processes

Cited by 29 publications

References 37 publications

Self‐correcting modifier‐adaptation strategy for batch‐to‐batch optimization based on batch‐wise unfolded PLS model

Self‐correcting modifier‐adaptation strategy for batch‐to‐batch optimization based on batch‐wise unfolded PLS model

Combining just‐in‐time modelling and batch‐wise unfolded PLS model for the derivative‐free batch‐to‐batch optimization

Constrained batch-to-batch optimal control for batch process based on kernel principal component regression model

Contact Info

Product

Resources

About