K-L法在结构模型降阶及辨识中的应用

APPLICATION OF K-L METHOD IN MODEL RANK REDUCTION AND IDENTIFICATION

模型辨识时,常需要对结构进行降阶处理,以减少辨识参数,降低计算难度.通过K-L(Karhunen-Loeve)特征值提取的方法对结构模型进行降阶,并通过最小二乘法辨识出降阶后系统的模态参数,可以大大减少辨识参数,降低计算复杂性;最后对一简支梁进行了数值仿真,得到一个3阶模型,辨识出简支梁的固有频率和模态阻尼比与解析解很接近.

In order to decrease the numbers of identification parameters and overcome difficulties in computation, the rank of structure should be reduced at first. In this paper, the rank of structure model is reduced by K-L(Karhunen-Loeve) eigenvalue extracting method. Modal parameters of the reduced model are identified by the least square method. The number of identification parameters and the computing complexity are greatly reduced. A simple girder is analyzed as an example. A three rank model is obtained. The identified inherent frequencies and the ratios of modal damping are close to the analytic result.