Robust and adaptive nonlinear control of electric motors and applications to robotic manipulators.

摘要

The application of robust adaptive nonlinear controllers to stepper motors and brushless DC motors using state and output feedback is explored in this dissertation. The control design methodology is based upon the earlier work of Jain [1]. Detailed models of the motors which incorporate observed physical phenomena (such as friction, magnetic saturation, cogging torque, eddy current damping effects etc.) are used to design controllers for variable reluctance (VR), permanent magnet (PM), dual-axis linear stepper (Sawyer) motors and for brushless DC (BLDC) motors, and the controllers offer robustness to both parametric and dynamic uncertainties as well as to bounded unmeasurable disturbances that may enter the system. The controllers developed are low order dynamic compensators as compared to most existing adaptive controllers developed for electric motors in the literature and therefore computationally simpler to implement in real-time. The tradeoff is loss of asymptotic tracking; however, the tracking error may be made arbitrarily small with appropriate choice of controller parameters.;Using only position measurements (i.e., the outputs), current-level controllers are developed for VR, PM and Sawyer motors which are robust to parametric and dynamic uncertainties and disturbances present in the mechanical dynamics of the system. The motivation for this design is the reduced cost and complexity of the system due to elimination of speed sensors. In this case, the control design is performed on a reduced order model of the plant, i.e., the mechanical subsystem; the current dynamics are neglected as they are much faster compared to the mechanical dynamics. Simulation results are provided to illustrate the efficacy of these controllers.;The connected problem of torque ripple reduction in stepper and brushless DC motors is addressed using a smooth adaptive variable structure control scheme. Torque ripple reduction essentially implies accurate control of the phase currents with a high bandwidth. Current-level controllers are designed using appropriate models for the motors that capture the torque ripple phenomenon present in these motors. The variable-structure type controllers require only one adaptation parameter (as compared to other schemes requiring adaptation to individual torque ripple harmonics) and are able to attenuate all torque ripple harmonics arising in the motor dynamics, as well as to uncertainties and disturbances (such as friction, unknown inertia, varying load torque etc.) entering the system. Furthermore, due to the smooth behavior of the controller, the phenomenon of chattering common to sliding mode controllers is avoided. Simulation results are provided to illustrate the efficacy of these controllers.