基于推广卡尔曼滤波算法的结构模型的参数识别

PARAMETER IDENTIFICATION OF STRUCTURAL MODELS BY MEANS OF EXTENDED KALMANIFILTER ALGORITHM

近年来,在结构工程领域中出现了许多参数识别技术,并将这些技术用于结构损伤程度的评价。但是目前多数结构识别方法都是模态参数识别。本文是基于推广卡尔曼滤波的算法,从结构模型的输入激励信号和输出的响应信号中,直接识别结构模型的物理参数。同时针对结构系统的特点,提出解耦的结构参数识别算法,大大降低了参数估计的计算量。在文中给出两个结构模型的几次振动台在地震波输入下识别的结果,并根据所识别的参数对结构模型破坏程度进行评阶,结果表明该方法对结构参数估计是一种有效的方法。

In recent years,some techniques for structural identification have been developed and applied to the damage evaluation of structures.However,most of techniques for parameter identification are those of modal parameter identification and other techni- ques for physical parameter identification need rather accurate initial estimates and a great quantity of computation time.In this paper,the algorithm of structural identifi- cation by means of Extended Kalman Filter(SIEKA)is proposed.Physical parameters and inner states of structures can be estimated from the input and responses by this mothod.Compared with other methods,the SIEKA algorithm is good at identifying physical parameters and states directly and dynamically. This paper,also introduces the principle and realization of SIEKA.For general structures,the idea of structural coupling and decouling is discussed.From the idea, decouled SIEKA algorithm for the widely used chained-multi-degree-of-freedom (CMDOF)structures is proposed in order to reduce the computation time. According to the input and response measurements of shaking table tests,the results of SIEKA's application to parameter estimation and damage evaluation of two CMDOF structure models are gives.It is shown that SIEKA is an effective algorithm for parameter estimaion and damage evalution of stuctures.