Batch processes are the preferred choice for manufacturing high-value-added products. The control performances of key process variables are critical to the product quality and quality consistency of batch processes. Most of the current control algorithms were originally developed for continuous processes. In comparison to continuous processes, batch processes have their own natures: repetive operation, two-time dimensional dynamics (within-batch and batch-to-batch dynamics) and multi-phase. To ensure good control performance, control system design and analysis must be conducted in harmony with the natures of the processes.

With such motivations, batch process control was studied systematically by fully exploring their features stated above. First, iterative learning estimator (...[ Read more ]

Batch processes are the preferred choice for manufacturing high-value-added products. The control performances of key process variables are critical to the product quality and quality consistency of batch processes. Most of the current control algorithms were originally developed for continuous processes. In comparison to continuous processes, batch processes have their own natures: repetive operation, two-time dimensional dynamics (within-batch and batch-to-batch dynamics) and multi-phase. To ensure good control performance, control system design and analysis must be conducted in harmony with the natures of the processes.

With such motivations, batch process control was studied systematically by fully exploring their features stated above. First, iterative learning estimator (ILE) with iterative learning control (ILC) was proposed for position control by utilizing the repetitive nature of batch processes. Second, the ILC has been integrated into the prediction model of dynamic matrix control, which leads to the 2 dimensional dynamic matrix control (2D-DMC). It is an integration of optimal feed-forward control and feedback control by using the repetitive and two-time dimensional dynamics (2D) nature of batch processes. Moreover, for systems with measurable states, an optimal guaranteed cost control scheme was developed via a robust H_∞ 2D controller for batch processes in an LMI framework. Last but not least, based on the proposed 2D-DMC, 2D hybrid dynamic models comprised of 2D models and general hybrid models concerning the 2D and multi-phase natures of batch processes were designed. With the 2D hybrid models, different design philosophies can be adopted for control algorithms design and tuning.

The modeling and control algorithms developed were tested on the MATLAB Simulink as well as an industrial-sized injection molding machine (typical batch process equipment). The control performance has been improved significantly in both the simulation and experimental test. The successful completion of this study not only addresses these important academic issues, but also provides a control method harmonious to the characteristics of batch processes.