Robust-optimal control of a servo motor

This work develops a frequency domain loop shaping design strategy based on linear quadratic gaussian controllers with loop transfer recovery (LQG/LTR). The design strategy presented is an integrated design synthesis procedure which is conducive to educated trade-off management. The controller design specification are expressed in terms of loop shaping bounds for load dynamic and model uncertainties, command response times and disturbance response times. The design philosophy is to first and foremost satisfy the stability constraints subject to the model variation specification and then to satisfy the time domain design specifications. The LQG/LTR controller is synthesised in discrete time, the resulting design is a Robust-Optimal design. Asymptotic disturbance rejection is achieved by using a disturbance estimate feedforward technique. Within the servo control industry two position feedback sensors are commonly used, namely the resolver and the optical encoder. In this work the controllers designed are applicable to both types of feedback sensors. Particular attention is given to the resolver sensor and corresponding estimator design. Often the resolver sensor is favoured because of its robust construction which is particularly suited for industrial environments. In this work a modified Extended Kalman estimator is developed which eliminates the need for a resolver to digital converter. A steady state solution for the Extended Kalman estimator is developed which provides a numerically robust and efficient estimator design. Finite manufacturing process and demodulation circuitry can often distort the resolver's outputs, these distortions introduce amplitude, bias and phase discrepancies onto the resolver channels which subsequently introduce velocity ripple into the closed loop system. In this work a parameter estimation scheme is developed which estimates and corrects the imperfections of the sensor interface