The objective of this research is to study the feasibility of using neural networks as the basis for intelligent and robust control of flexible structures. Currently, implementation of modern control systems is based on sequential processing of state equations as opposed to the massively parallel processing architecture (neural network) of the human brain. As systems become more complex, the ability to control them by traditional sequential control methods becomes limited. Parallel processing architecture is inherent in the recently developed technology of neural networks. It offers the potential for addressing the limitations associated with traditional sequential control methods. There are three potentially powerful characteristics associated with neural networks. They are: (1) the ability to learn and make decisions based on an optimization criteria; (2) the ability to perform a recognition task under uncertain conditions and with incomplete information: and (3) a high processing speed. This project will study whether these characteristics can be utilized in solving a flexible structure control problem. The three phases of the research include development of a neural space representation for complex systems, formulation of the control system structure in terms of the neural network structure, and implementation of the results on the control of a flexible beam.
