On-Line Adaptive Optimal Control

The development of algorithms which will optimize the performance of a process in real time has been receiving considerable interest. The most robust of these algorithms use process identification techniques developed for adaptive control technology. The steady state portion of the model is then used to determine changes which should be made in the system inputs in order to drive the process toward the point at which it operates at its optimum. Currently these algorithms use only the steady-state portion of an identified dynamic model. This paper involves the use of optimal control and the identified dynamic model to shorten the time required for optimization. The optimal trajectory is recalculated and the process model is updated every sampling period.