Empirical frequency-domain optimal parameter estimate for black-box processes

Most of the previous signal processing identification results have been achieved using either time-domain or frequency-domain algorithms. In this study, the two methods were combined to create a novel identification algorithm called the empirical frequency-domain optimal parameter (EFOP) estimate and the recursive EFOP algorithm for common linear stochastic systems disturbed with noise. A general prediction error criterion was introduced in the time-domain estimate. By minimizing the frequency-domain estimate, some general prediction error criteria were constructed for Black-box models. Then, the parameter estimation was obtained by minimizing the general prediction error criterion. This method theoretically provides the globally optimum frequency-domain estimate of the model. It has advantages in anti-disturbance performance, and can precisely identify a model with fewer sample numbers. Lastly, some simulations were carried out to demonstrate the validity of the new method.