Adaptive Control of Nonlinear Mechanical Systems

• Main Authors: Chen-yu Kai, 蓋震宇
• Other Authors: An-Chyau Huang
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/30416083069954322204

博士 === 國立臺灣科技大學 === 機械工程系 === 101 === This dissertation proposes several adaptive control design schemes for special classes of nonlinear mechanical systems. In the first part a new paradigm is proposed to construct a regressor-free adaptive controller for robot manipulators without Slotine and Li’s modification. Similar to the traditional adaptive strategies for robot manipulators, the regressor-free adaptive controller design also requires applying Slotine and Li’s modification to avoid the feedback of joint accelerations. In the new paradigm, the joint acceleration vector is lumped into an unknown time-varying function and the function approximation technique is utilized to cover its effect; therefore, the new paradigm needs neither the regressor matrix nor acceleration feedback. Since the new paradigm can easily be used to design the adaptive controller, it is obviously applicable to other control schemes. Therefore, the same paradigm is going to be applied to compliant motion and optimal control problem. In the second part of this dissertation, vibration suppression control for mechanical systems is considered. Conventional vibration absorber is valid when the external excitation frequency falls into a neighborhood of a specific value. It is not feasible for vibration suppression applications subject to frequency-varying excitations. In this dissertation, an adaptive controller is proposed to learn the disturbance spectrum in real-time so that output vibration suppression can be achieved regardless of the variations in the excitation frequency. In the third part, we consider adaptive controller of brushless DC motors without model reduction. Almost all controllers designed for BLDCM are based on reduced models where a first order dynamics is regarded as the internal model. This way a second order dynamics is considered with considerable simplification in the controller design. Since the reduced model may not reflect the entire dynamics of the motor, some deterioration in the control outcome can be observed. To improve the performance, this dissertation considers the adaptive control of a BLDCM without using reduced model. The last part of this dissertation proposes a new linearization algorithm for rate-dependent nonlinearities. The rate-dependent nonlinearity presents different dynamics in response to variations of the input signal frequency. Both the Jacobian linearization and feedback linearization strategies cannot give proper performance for systems with these nonlinearities. The new linearization algorithm is proposed for rate-dependent nonlinearities where a Fourier series based function approximator in feedback configuration is designed to cover the effect of the nonlinearity such that a desired linearity between the input and output signals can be realized. The proposed strategy is so general that the nonlinearity is allowed to be uncertain. All of these adaptive control schemes, the Lyapunov-like analysis is used to verify the closed loop stability and ensure boundedness of internal signals. Simulation and experimental cases are given to show efficacy of the proposed schemes.