数据同化
计算机科学
补偿(心理学)
控制理论(社会学)
算法
实时计算
数据收集
工程类
数学优化
数值模型
作者
Keita Uemura,Ryota TSUTABA,Sho Kameda,Yuya Morishita,Yoshikazu Takahashi
标识
DOI:10.1080/15732479.2025.2575024
摘要
Real-time hybrid simulation (RTHS) faces the challenge of actuator time delays, which disrupt the correlation between experimentally measured forces and displacement responses computed by numerical analysis, thereby reducing the accuracy of the RTHS. This study proposes a concept of applying data assimilation method for delay compensation in RTHS. The proposed method compensates for measured forces by utilising a simplified nonlinear model (compensation model) of the test specimen, constructed from prior information about the specimen. This model is continuously updated using a data assimilation method based on the unscented Kalman filter. The proposed method is distinctive in that it utilises the updated numerical model solely to correct force errors resulting from actuator time delays. The effectiveness of the proposed delay compensation method was evaluated through virtual RTHS. The results demonstrated that the proposed method successfully compensated for significant actuator time delays ranging from 100 ms to 200 ms. Specifically, the normalised root mean square error in the virtual specimen’s response displacement was reduced to less than one-fifth of that observed with the commonly used ATS (adaptive time series compensation) method. Future research should explore applying the method to RTHS with real actuators and more pronounced nonlinear experimental specimens for feasibility.
科研通智能强力驱动
Strongly Powered by AbleSci AI