A Method of Noise Model Identification Based On M-Series

It is necessary that the signal-to-noise ratio of the measured output signals must be large enough when we use the methods of identification to establish the mathematical model of the system. If the power of the noise in the output signals is too large, the model established through system identification will not reflect the fact, and the controller based on the model will not get expectant effect. In this application, we need establish the model of noise, and preprocess the output signals according to the model. This paper provided a least-square method to identify the noise model using M series as the input signal of the noise model.