On-line damage identification and motion control of adaptive structures: An intelligent control approach

In this dissertation, an on-line damage identification technique and a motion controller for the adaptive structure are developed under the scope of an intelligent control framework. The computational intelligence techniques, especially neural networks and fuzzy logic control, are studied on their application to the adaptive truss structure.; The adaptive structure has the ability to change the configuration and/or physical properties amenable to the varying environment or to accomplish various mission requirements. This dissertation first overviews an intelligent control framework specified for the adaptive structures. The intelligent control system consists of three hierarchical levels, from up to down, Management and Organisation level, Coordination level, and Execution level. The elementary classification of the levels and the responsibilities of major functions in each level are described in this study.; The work then investigates two major functions in the Execution level: structural damage detection and identification, and motion control. A neural-network-based damage identification technique which is able to detect the damage location accurately and then to identify the damage severity with reasonable precision is proposed. The new architecture of neural network provides several advantages over conventional approaches: first, this technique only needs as few as possible modal data, i.e. eigenvalues and normalised mode shape, to identify structural damages; second, it can detect multiple damages case; third, the process can be realized on-line, moreover, the damage location detection can be done in real-time.; Next, a fuzzy logic controller is used for the motion control of the adaptive truss structure. A systematical procedure to design fuzzy controller is proposed. The fuzzy rules can be constructed by referring the kinematics of the structure, thus omitting conventional trial-and-error approach. The controller was tested on several different tasks such as rendezvous/docking, target pursuing, and trajectory tracking. The results show the control scheme is effective and reliable. The effects of the performance by varying design parameters are also investigated. The general guidelines for selecting the design parameters are addressed.