Time delay observer

Abstract

Motivated by the idea of the time delay estimation used in the Time Delay Control (TDC), Time Delay Observer (TDO) is constructed firstly for nonlinear single-input single-output (SISO) plant in phase variable form in Chapter 1. Owing to this idea, TDO does not require a model, and hence is easy to design for nonlinear systems in the phase variable form, even when the plant dynamics is not well known. Not in-cluding a model, TDO turned out to be robust to parameter variations and computationally efficient; yet achieved better accuracy. For this observer, the convergence property and the sensitivity to sensor noise was analyzed. Then, in a simulation with a nonlinear plant, TDO achieved an accuracy level similar to sliding mode observer and better than a Kalman filter. Chapter 3 extends the TDO for nonlinear SISO plants in the phase variable form to MIMO plants. The extended TDO reconstructs the states of the plant expressed in the generalized observability canonical form (GOBCF), yet requiring neither the transformation of a plant, nor the real time computation of system nonlinearity. Consequently, when the TDO is used for several control schemes based on the transformed coordinates, the observer turned out to be computationally efficient and easy to design for nonlinear MIMO plants. In a simulation of a two-link manipulator with flexible joints, the control performances using TDO appeared to be similar to those using actual states and superior to those using numerical differentiation. Chapter 4 presents input/output linearization (IOL) method using TDC and TDO. This method enables the IOL method to plants when all the states of plant are not measurable or the measured plant output is very noisy. The designed control system requires neither an accurate plant model nor the real time computation of plant nonlinearity. Consequently, the proposed control algorithm turned out to be computationally efficient and easy to design for nonlinear plants. In a simulation fo...