Optimal integrator-less tracking controller for uncertain systems

Tracking controllers are designed to drive the system on the desired trajectory with minimum error. In real life, all systems have uncertainties such as friction, modeling inaccuracies or damage that cause the performance to deviate from the ideal response. This research is aimed at addressing the tracking problem in the optimal context and a generalized method to handle the uncertainties is proposed. In the first paper, an optimal tracking controller without an integral component is formulated for a linear system. This design can handle time varying, non-zero references and effectively alleviates the risk of \u27wind up\u27. A neural network based observer is used in the estimation of system uncertainties and this information is applied to adaptively redefine the control. The theoretical findings were demonstrated by simulation. The second paper applies the control scheme in paper I to the linear axes of a mini CNC machine for the contour control of two orthogonal axes. This process is susceptible to dynamically varying nonlinear friction, which is difficult to model. The experimental results are presented and reflect the excellent tracking performance without the use of integral control. It was found that the controller was robust to significant variation in the model gains. A model-following controller for high-performance aerospace applications based on the optimal tracking controller is put forth in the third paper. The objective is to make the system follow a dynamic model with ideal characteristics, while handling the nonlinearities induced by changing operating conditions or aggressive flight regimes. Simulations studies showed that the controller can adapt to the state and control dependent nonlinearity without inducing unnecessary oscillations --Abstract, page iv