摘要
扩展卡尔曼滤波(extended Kalman filter,简称EKF)方法常用于结构参数识别,但存在对滤波参数敏感等局限性,需大量试错来寻找最优噪声方差参数。针对此问题,推导了基于残差的协方差匹配公式。首先,通过滑动窗口法或遗忘因子法自适应更新匹配测量噪声方差,实现了基于EKF的自适应识别结构参数;其次,以一个3层Duffing型非线性剪切框架为例来验证方法的有效性,并进行了参数鲁棒性分析。结果表明:滑动窗口法和遗忘因子法均能很好地估计测量噪声方差,识别效果和收敛速度接近;与非自适应EKF方法相比,自适应EKF方法对噪声方差的初始取值不敏感,具有很强的鲁棒性。
The extended Kalman filter(EKF)method is commonly used for structural parameter identification.The EKF has limitations such as sensitivity to filtering parameters,and requires the trial and error method to find the optimal noise variance parameter.In this paper,a residual-based covariance matching formula is derived,and the covariance matrix of measurement noise can be adaptively updated by either the sliding window method or the forgetting factor method,and the adaptive identification of structural parameters based on the extended Kalman filter is realized.A three-storey Duffing-type nonlinear shearing frame is taken to verify the effectiveness of the method,and the parameter robustness analysis is carried out.The results show that both the sliding window method and the forgetting factor method can estimate the measurement noise variance well,and the recognition effect and convergence speed are close;Compared to the non-adaptive EKF method,the adaptive EKF method is insensitive to the initial value of the noise variance and has strong robustness.
作者
万华平
马强
欧一鸿
张文杰
周家伟
陈伟刚
WAN Huaping;MA Qiang;OU Yihong;ZHANG Wenjie;ZHOU Jiawei;CHEN Weigang(College of Civil Engineering and Architecture,Zhejiang University Hangzhou,310058,China;Center for Balance Architecture,Zhejiang University Hangzhou,310028,China;Architectural Design&Research Institute of Zhejiang University Co.,Ltd.Hangzhou,310028,China;Zhejiang Southeast Space Frame Co.,Ltd.Hangzhou,311209,China)
出处
《振动．测试与诊断》
EI
CSCD
北大核心
2024年第6期1082-1089,1244,共9页
Journal of Vibration,Measurement & Diagnosis
基金
国家重点研发计划资助项目(2021YFF0501001)
浙江省重点研发计划资助项目(2021C03154)
国家自然科学基金资助项目(51878235)。
关键词
结构参数识别
鲁棒性
自适应扩展卡尔曼滤波
滑动窗口法
遗忘因子法
structural parameter identification
robustness
adaptive extended Kalman filter
sliding window method
forgetting factor method
