Intelligent Control Using Active-Adaptive Information Sources

In this paper, we report preliminary results of a framework within which to view intelligent controllers. Our premise is that an intelligent control system differs from a traditional control approach primarily in its usage of information. Unlike a traditional feedback control system, an intelligent controller uses a variety of information sources in a flexible form which are updated on-line as and when necessary. Motivated by the human decision process which activates various new information sources on-line in order to resolve difficult and confusing situations, we propose a framework wherein the controller can update the vector information sources based on knowledge and experience gained on-line, even as the system is in operation, and appropriately reorganize the control structure automatically. In particular, we shall focus on the contigency situations, examine candidates for contingent variables that needed to be added to the existing outputs, and propose a measure of persistent excitation to evaluate the information content of each contingent variable.