A batch process management framework: Domain-specific, design pattern and software component based approach: Dissertation

Requirements for product and production quality and variability, as well as the needs for the efficient use of production equipment, have emphasised the benefits of batch production in the process industries. The resulting complexity of batch control has, however, been a challenge to control engineers. Emerging batch standards and software component technologies have now made it possible to design flexible, distributed, and integrated batch automation concepts to satisfy the requirements. The batch control domain was studied in this thesis in terms of domain standardisation, existing batch control systems, and related research approaches. The applicable information technology, object-oriented software component frameworks and multi-agency, was surveyed and evaluated. Guidelines for deploying generic software design patterns in designing domain-specific frameworks, were adapted. An experimental batch process management framework, was developed to fulfil the aforementioned flexibility, distribution and integration needs for batch automation. It also demonstrates reusability by the so-called calling framework architectural style as well as internal and external component interfaces. Framework components may be easily parametrized and replaced by customised versions. Additionally, the framework can be integrated with other systems by using component technology. For some problem specific needs of local decision-making and interaction, enhancement of component frameworks may be needed. No applicable design patterns were found for this kind of design issue. Since the design problem is recurrent, a generic design pattern, Agentified Component, was developed and experimented with within the framework of this thesis. The approach retains the deterministic nature of the framework, important in the automation domain, but simultaneously introduces the possibility of solving specific problems using a knowledge-based approach.