Batch to batch optimal control based on multiinput multioutput adaptive hinging hyperplanes prediction and Kalman filter correction

Abstract

Summary A batch to batch optimal control strategy based on multiinput multioutput adaptive hinging hyperplanes (MIMO AHH) prediction and Kalman filter correction is proposed for the products quality control of the batch process. The model of AHH is one kind of piecewise linear models and is extended to the MIMO case in this article. The MIMO AHH is then used to develop the predictive model of the batch process. Due to the model‐plant mismatch and unknown disturbances, the optimal control policy calculated based on the MIMO AHH predictive model may not be optimal when applied to the true process. The Kalman filter is then utilized to correct the predictions of the current batch by considering the information of former batches. The effectiveness of the proposed strategy is verified through the simulation of a styrene batch polymerization reactor.

State Estimators in Soft Sensing and Sensor Fusion for Sustainable Manufacturing

“…Luo et al [55] studied batch-to-batch polymerisation and proposed an adaptive hinging hyperplane (AHH) model for the process, which is a type of piecewise linear model for nonlinear systems. A MIMO (multi-input multi-output) model was developed to predict the process behaviour.…”

Tube‐based batch model predictive control for polystyrene polymerization reaction process

“…4 One of the important objectives of the polystyrene polymerization reaction process is to ensure that the enduse polystyrene products have the desired properties for a specific application, such as the molecular weight distribution, chain lengths, and monomer conversion. For this purpose, various advanced control techniques, including proportional-integral-derivative (PID) control, 5 fuzzy control (FC), 6 iterative learning control (ILC), [7][8][9] and model predictive control (MPC), 10,11 have been implemented to improve the product quality of polystyrene polymerization reaction process. Some typical works on this topic in the last two decades are summarized in Table 1.…”

“…A batch-to-batch ILC method based on a time-varying perturbation model was presented by Xiong et al 9 Jia et al 7 used a neuro-fuzzy model to establish the ILC control and provided a thorough convergence analysis. An adaptive hinging hyperplanes model and Kalman filter technology were combined by Luo et al 8 to propose an optimal ILC design approach for a batch polystyrene polymerization reactor with multiple inputs and outputs. The available ILC strategies may not always achieve a good tracking performance if real-time disturbances affect the polystyrene polymerization reaction process because the ILC method is an open-loop control strategy and lacks real-time feedback information.…”