Identification of Stochastic Process in MATLAB

The system identification toolbox in MATLAB has been successfully used to compare model identification of a first order system subjected to high and low disturbances. The model structures used are FIR, ARX, AMX, OE and BJ. The obtained Model was validated using data generated from the actual process. It shows that the more the variance of the noise input into the system, the more difficult it is for the model identified to reproduce that validation data obtained from process response. Also when the measurement noise has zero mean and low variance, the effect on the steady state gain and other process parameters is negligible