Backstepping control techniques for a class of mechatronic systems.

摘要

This dissertation investigates the modeling and nonlinear control of several electromechanical systems including a programmable active dynamometer, an active magnetic bearing (AMB), a self-bearing switched reluctance motor (SBSR), and a super clean direct-drive robot. The control designs are based on backstepping techniques to yield voltage level controllers. First, a computer-controlled active dynamometer system consisting of permanent magnet brushless DC motors is proposed and the model-based nonlinear controller is designed. Secondarily, two control designs are proposed for the AMB system. One is model-based nonlinear control. Another is adaptive control to deal with unbalance and unknown static load change. Then, the idea of the SBSR motor is proposed and its modeling, commutation, and control are discussed. Next, a super clean direct-drive robot is proposed and its modeling and control are discussed. All the model-based controllers for the above systems yield global exponential stability of the closed-loop systems. The adaptive control for the AMB system guarantees, for the first time in the literature of Lyapunov-type non-linear adaptive control, exponential stability and parameter convergence to the true value. Finally, as part of the future work, a new problem about nonlinear control design allowing computing time is formulated and a nonlinear optimal control approach is proposed as an initial attack to this difficult problem. A linear case study is presented for this problem. At the end of the dissertation, the completed work is summarized and further investigation with the emphasis on the modeling, commutation, and control of the SBSR motor to consider the regular rotor geometry and mutual inductances as well as saturation using truncated Fourier series approximation, polynomial approximation, system identification techniques, and neural networks, is suggested. Future research topics are also outlined.