Nonlinear system identification and control using fast orthogonal search

This thesis is concerned with system identification and control of nonlinear systems. In order to control a nonlinear system, either a linear or a nonlinear approximation can be made. The former allows for simpler linear control methods to be used, however if the nonlinearity of the system is significant then those methods may not result in sufficiently accurate control. In the latter case, a better model and control of the system generally result, however nonlinear control methods are typically very difficult to use and computationally very intensive.; A straightforward, simple-to-use method of linearizing nonlinear systems will be described, which allows for simple (linear) control methods can be used. An orthogonal system identification method known as fast orthogonal search (FOS) is used for this linearization.; The control concept is derived from the method of inverse dynamic control (IDC). The novel combination of this with the FOS method of system identification results in very efficient and effective methods of control. Three variants of the method are presented: a nonlinear cancellation scheme and two methods of system inversion.; The effectiveness of these methods of control is tested on chaotic systems, chemical process control and mechanical system control problems. The last and most complex problem is a multiple-input, multiple-output mechanical machine in which real data are used for identification of the model.; The results compare very favourably to other published results and methods. The methods presented here are also compared to each other.