An ARMA robust system identification using a generalized lp norm estimation algorithm

The parameter estimation of ARMA systems with noisy output data is considered. The system output is corrupted by measurement noise, and the noise distribution is assumed to be unknown. A generalized l_p norm iterative estimation algorithm (1<p<∞) is proposed to approach the maximum likelihood estimate of system parameters according to the noise distribution. Since the exponent p of l_p norm estimation algorithm is sensitive to the noise distribution, based on the sample kurtosis of the residual, an adequate exponent p of l_p norm estimation algorithm can be selected to achieve the efficient parameter estimation at each iteration step. Finally, several simulation results are presented to illustrate the proposed l_p norm estimation algorithm; and the authors find that the proposed generalized l_p norm estimation algorithm offers significant advantage to robust system parameter estimation problem with unknown noise distribution