Recursive prediction error algorithm for general full structure model
Recursive prediction error algorithm for general full structure model
Summary: The quasi-Newton recursive prediction error identification method has properties such as fast convergence, unbiasedness, and numerical robustness. \textit{L. Ljung} [Automatica 17, 89-99 (1981; Zbl 0451.93064)] has proposed a fast calculation algorithm for the gain matrix in a full structure model with the same orders of polynomials. However, it seems that this algorithm is not applicable in practical use, because it can only be employed under a special assumption. In this paper, a fast calculation algorithm for the gain matrix L(t) with different orders of polynomials is deduced. The recursive prediction error algorithm can be easily used in on-line signal processing.
numerical robustness, fast convergence, quasi-Newton recursive prediction error identification, on-line signal processing, System identification, Computational methods in systems theory
numerical robustness, fast convergence, quasi-Newton recursive prediction error identification, on-line signal processing, System identification, Computational methods in systems theory