Identification of systems from multirate data

Abstract

Advanced process control implementation in the chemical industry relies heavily on process models obtained from experimental plant data. System identification techniques are routinely employed for this task. However, it is often assumed that the plant input and output measurements are sampled regularly and at the same rate. In many processes, the controlled variables (mainly composition, melt flow index) are sampled very slowly and at irregular intervals while the other variables (flow rates, pressures, temperatures) are measured frequently and regularly. Such multiple rates of measurement naturally leads to multirate data sets. The identification of models that can be used for controller design using multirate data will be the focus in this work. The works focuses on some of the methods for multirate system identification of linear system and nonlinear system, associated issues and applications.