Development of an analysis framework for measurement-based dynamic optimization

点击翻译按钮获取中文摘要

Development of science consists of inventing new ideas and implementing them efficiently. Once a new process has been developed, optimizing its operation is essential for its commercialization. Also, the importance of optimization resurfaces, when an existing process faces tough economical and environmental challenges. Usually, optimization is carried out numerically based on a mathematical model of the process. However, these models are not always accurate and the numerical solution so obtained does not correspond to the optimal operation. So, the context of the current project is measurement-based optimization, where measurements taken from the process are used for optimization, instead of a mathematical model. Among measurement-based optimization techniques, there has been considerable effort in the domain of unconstrained static optimization, where the best operating point without constraints is determined.  However, the proposed research addresses the more complex and realistic problem of constrained dynamic optimization, where optimal time profiles are sought in the presence of constraints both during the operation and on the final product specifications. The objective of the research is to provide an analytical framework for the constrained dynamic optimization problem based on measurements. This framework will lead to the development of efficient algorithms, which can be employed not only for operational problems, like improving the productivity of an industrial unit, but eventually also to scheduling and supply chain management problems.