Author: HAN Chenye | Time: 2021-01-10 | Counts: |
doi:10.16186/j.cnki.1673-9787.2019090100
Received:2019/09/25
Revised:2020/02/21
Published:2021/01/15
Parameter estimation of unknown order Wiener system based onmulti-innovation algorithm
HAN Chenye
School of Information Technology, Hebei Polytechnic Institute, Shijiazhuang 050091 , Hebei, China
Abstract:Aiming at the over-parameter problem of identification model forthe unknown order Wiener system and the low accuracy of the least-squares method, a determinant ratio method and the multi-innovation least squares approach based on decomposition technique were proposed. Firstly, data matrix was constructed from system data, and the order of the system was estimated by determinant ratio method. Secondly, the linear model was substituted into the special term of the nonlinear model by using the decomposition technique, and an estimation model of parameter separation of linear and nonlinear model was established, which reduced the computation burden of the algorithm. Then, a reference model was developed to handle the internal signal, which transformed the unknown internal signal into the indirectly measurable signal. Finally, scalar innovation was revised to multi-innovation through the usage of the certain length, which improved the performance of scheme. Furthermore, the influence of different noise and different innovation length on the proposed algorithm were analyzed. The simulation results showed that the proposed estimation scheme outperformed the recursive least squares method in estimation accuracy and convergence rate.
Key words:parameter estimation;unknown order Wiener system;decomposition technique;multi-innovation least square;reference model