利用快速子空间跟踪的时变模态参数识别

Parameter identification of a time-varying system with a fast subspace tracking method

提出利用一种快速子空间跟踪技术(API)解决基础激励下时变系统的模态参数识别问题。利用指数窗和截断窗更新状态向量及其压缩向量的互相关矩阵,通过正交化确保特征权重矩阵的正交性,推导特征权重矩阵的递推公式。数值仿真时在API方法中加入数据更新速率,利用结构在基础激励下的动力学方程,通过数值方法获得时变系统随机振动响应,仿真结果表明该方法既能保证识别精度,又能减少计算量。

A subspace tracking method with API is proposed to solve parameter identification of a time-varying system under basement excitation. The cross correlation matrix of state vector and compressed data vector is updated by exponential window and truncated window. The orthogonality of the characteristic weight matrix is guaranteed by the orthogonalization step and recurrence formula of characteristic weight matrix is derived. The data refreshing rate is added into API method in numerical simulation and the vibration response of a time-varying system is obtained by dynamical equation under basement excitation. The simulation results show that API method offers good identification accuracy and less computation.