A New Parameter Estimation Algorithm for CARMA Models

A modified recursive maximum likelihood (RML) parameter estimation algorithm is presented in this paper. The white noise estimations obtained by fitting the CARMA model to the high-order controlled autoregressive(CAR) model using the recursive least squares (RLS) method. Using these white noise estimations into RML parameter estimation algorithm, which can solve the problem that parameters estimation becomes slow when the control parameters and noise parameter are tightly coupled. The modified RML parameter estimation algorithm has many advantages such as simple algorithm, small calculation amount, and high identification precision, good convergence. It can be used for on-line identification and real-time data processing, with theoretical significance and practical value.