A comparative study of nonlinear observers and control design strategies.

摘要

Conventional observers are based on an accurate mathematical model of the plant, which limits their applications in the real world where the accurate model may not be available. The Extended State Observer (ESO) is a novel, model-independent observer. It has been successfully applied to many practical applications: such as hard disk computer drives, web tension regulation, DC-DC converters, and industrial servo systems.; This research focuses on the comparative study of different observers and observer based control strategies. The purpose is to find a set of practical solutions that make control design and tuning easy and effective. Three observer methods, including the ESO, the High Gain Observer (HGO), and the Sliding Mode Observer (SMO) are investigated. The simulation and experimental results of motion systems show that ESO has a better overall performance. Active disturbance rejection control (ADRC) is an ESO-based control strategy. With a novel, optimization-tuning method, linear ADRC achieves effective results when dealing with large inertia, friction variations and torque disturbance. A simulation of an active, magnetic-bearing benchmark problem is used to test the method. Based on the Popov criterion, a stability analysis of second order ADRC is investigated.