多变量AR建模方法在工作模态参数辨识中的应用

对于一些大型工程结构,有时很难甚至无法测量结构的输入信号,只能利用实测的响应数据进行工作模态识别.针对这一实际工程情况,首先利用工况下结构的响应数据作相关分析,继而对相关分析后得到的多组数据构造多变量自回归(AR)模型,依据逐步最小二乘法和准则函数优化思想确定模型阶次和估计模型参数,并由模型参数矩阵的特征值分解确定结构的工作模态参数,最后利用TH6350大型镗铣加工中心主轴系统进行试验验证,结果对比表明,这种方法不仅能够有效提取结构动态特性参数,而且还能提供有关激励和结构振型的信息.

Application of Multivariate Autoregressive Modeling in Operational Modal Parameters Identification

Multivariate Autoregressive modeling is a new method for structural modal analysis under working condition.The process of modeling only depends on the output response signal of system.It is special for the modal analysis of system which is difficult or almost no methods to test excitation.This method is based on multivariate autoregressive models,chooses models order by modeling optimized thinking and estimates parameters by the Step-wise Least Square Algorithm.The structural modal parameters are estimated by the eigendecomposation of the parameters matrix of AR models.The method is successfully applied to analyze dynamic characteristics of spindle system of TH6350 bed machining center and the modal parameters of system are extracted under working condition.Moreover,this method can also offer some information of excitation and modal shape of structure.It has presented a believable technique for the parameters identification of linear system dynamic characteristic under operational condition.