|
| Research on the identification of hysteresis model parameters based on the modified unscented Kalman filter algorithm |
| Received:August 27, 2022 Revised:November 30, 2022 |
| View Full Text View/Add Comment Download reader |
| DOI：10.7511/jslx20220827004 |
| KeyWord:hysteresis behavior BWBN model modified UKF algorithm state estimation parameter identification |
| Author | Institution |
| 夏运达 |
上海市政工程设计研究总院集团, 上海 |
|
| Hits: 344 |
| Download times: 227 |
| Abstract: |
| The structure of reinforced concrete bridge pier exhibits hysteresis behavior when subject to cyclic loadings produced by earthquakes.To describe the hysteresis behavior,researchers have proposed various hysteresis models,among which BWBN (Bouc-Wen-Baber-Noori) model can describe typical characteristics including strength degradation,stiffness degradation and the pinching effect.Besides,UKF (unscented Kalman filter) algorithm is an efficient method to identify parameters of BWBN model but its identification process is restrictive when the deviation between the initial and real values of parameters is big enough and lacking of overall estimation for the system.In the paper,the rule of generating sample points is modified and the modified UKF algorithm is proposed .The numerical simulation results show that the average error between estimated and real value of parameters is 1.51% and the maximum error is 4% with no noise;the average error is 5.43% and the maximum error is 18% with 2% RMS(root mean square) Gaussian white noise;the average error is 8.9% and the maximum error is 26% or 22% with 5%RMS Gaussian white noise.The modified UKF algorithm is more accurate and robust to identify the state estimation of a non-linear hysteresis system and parameters of a BWBN model. |