Control of non-smooth and nonlinear dynamic systems.

摘要

Control of nonlinear non-smooth system is an important research topic. In this thesis, two typical nonlinear characteristics—deadzone and time-delay are investigated. Deadzone is common in electrical and mechanical systems. We have developed a sliding mode control explicitly accounting for the hardware limitation of mechanical actuators with deadzone, resulting a continuous control. Simulations are presented to demonstrate the effectiveness of the proposed controls.; Model predictive control is an effective method for systems with time-delay. In our implementation of MPC, we investigate the benefit of predicting the future reference based on its past history. Simulation results show the advantage of the prediction for slow-varying signals.; We also present an application of semi-discretization method to stability analysis of PID feedback controls of linear systems with time delay. The method creates a mapping of the system response in a finite dimensional state space. Minimization of the largest absolute value of the eigenvalues of the mapping leads to optimal control gains. Numerical examples of both time-invariant and periodic linear systems are presented to demonstrate the method. We have found that the semi-discretization method provides accurate stability boundaries and performance contours in the parametric space of control gains, and offers an alternative to the classic design approaches of feedback controls.