Nonlinear internal model control based on local linear neural networks

Abstract

The internal model control (IMC) scheme has been widely applied in the field of process control. This is due to its simple and straightforward controller design procedure as well as its good disturbance rejection capabilities and robustness properties. So far, IMC has been mainly applied to linear processes. This paper discusses the extension of the IMC scheme to nonlinear processes based on local linear models where the properties of the linear design procedures can be exploited directly. The resulting controllers are comparable to gain-scheduled PI or PID controllers which are the standard controllers in process industry. In practice, the tuning of conventional PI or PID controllers can be very time-consuming. In this paper, the design effort of the nonlinear IMC and conventional controller design methods are discussed and the control results are compared by applying it to a Hammerstein process and nonlinear temperature control of a heat exchanger.