Eigen-Space Decomposition (ESD) Method for the Design of Internal Model Controller (IMC) from Noisy Input and Output Plant Data

A novel approach to design Internal Model Controller (IMC) is proposed in this paper directly from measured input and output plant data, which are assumed to be contaminated by measurement noise. In order to avoid the complicated structure - identification problem in most cases, two Finite Impulse Response (FIR) models are taken to represent the plant model and the internal model controller respectively. Taking account of measurement noise both in the plant input and its output, anESD based Total Least Squares(TLS) solution is applied for the unbiased Identification of the plant model and its inverse model, the latter constitutes the internal model controller according to the principle that the internal model controller approximates the inverse dynamics of the plant model. Simulations are given for a testification.