基于泛函序列时变自回归滑动平均模型的弹箭时变模态参数递推估计方法

A Time-varying Modal Parameter Recursive Estimation Method Based on FS-TARMA Model for Missile and Rocket

随着弹簧系统朝着大型化、高速化、智能化发展,飞行状态下弹箭系统的固有特性对整体结构的影响不可忽视。针对弹箭在飞行状态下的时变模态参数辨识问题进行研究,基于泛函序列时变自回归滑动平均(FS-TARMA)模型,提出一种时变模态参数的递推估计方法。该方法采用墨西哥帽小波基作为TARMA模型时变系数的空间基底,并借鉴于无结构化TARMA模型递推估计思想,将投影参数矩阵视为振动响应数据长度的变量,实现了投影参数矩阵的递推估计。通过有限单元法建立阿里安V号芯级运载火箭时变有限元模型,对所提方法进行验证。结果表明:递推辨识方法与传统批量算法相比,在辨识精度上,3阶模态频率辨识结果最大相对误差在5%以内;在计算效率上,递推辨识方法的计算时间缩短了9.38倍。

With the development of missile and rocket system towards large-scale,high-speed and intelligent,the influence of dynamic characteristics on the overall structure cannot be ignored.The problem of time-varying modal parameter estimation for missile and rocket duing operation is studied.A novel time-varying modal parameter recursive estimation method is proposed based on functional series time-dependent autoregressive moving-average(FS-TARMA)model.The proposed method uses wavelet function as the basis function of time-varying parameters in TARMA model,and the projection of parameter matrix is recursively estimated as it is treated as a variable of the length of vibration response data by using the recursive manner of unstructured TARMA model.The proposed method is validated by the core stage of Ariane V time-varying finite element model.The verification results demonstrate that the proposed recursive method has nearly same identification accuracy compared with batch estimation method,with the relative RSS error in three-order modal frequency being within 5%,and has superior computation efficiency as the computation time is shorten to 1/9.