Batch processes arc of considerable importance in the modern chemicals industry, especially in the productions of fine chemicals. Industrie 4.0 has facilitated renewed research interests in the batch process, thanks to its great flexibility to meet rapid-changing markets and capability to produce high-value-added goods. However, compared to continuous process, its unique dynamic characteristics have brought both opportunities and challenges to the process control community. Therefore, both the economic and academic values of the batch process haw motivated this work.

Within this thesis, the control, identification, estimation and optimization problems of batch process arc systemically analyzed. The primary objective was to exploit the repetitiveness of batch process in order to improve the performance of the aforementioned three problems. This study proposes a weld-line positioning controller for the bi-injection molding process. An online identification algorithm incorporating priori process knowledge of pole and zero positions has been devised to yield smooth and accurate process dynamic models, which allows parallel computation. Two averaging based identification methods have been developed to handle the inter-batch dynamics drift. For the stale estimation problems, a robust Kalman filter with iterative learning mechanism has been designed to deal with model mismatches as well as to refine estimates from iteration to iteration. An iterative learning extremum seeking control method has been proposed to optimize the batch process operating line. Theoretical analysis tools have been developed to inspect the four methods. Numerical experiments act as testimonies to the effectiveness of the proposed methods.