On-line control model validation through simulation

The use of digital signal processors (DSP) in the evaluation of control models and in the detection of process and sensor faults is discussed. Based on measured process inputs a DSP can be used to simulate the process. A multivariate statistical analysis is applied on the difference between the true outputs and the simulated outputs. The detection and identification methods for model invalidity are discussed. Some important pulp and paper applications are outlined.<
>