连续转速状态下卫星反作用飞轮微振动参数识别

Identification method of micro-vibration parameters of satellite reaction wheelsunder continuous rotating speed

由于制造和装配误差,反作用飞轮产生的谐波干扰是导致光学遥感卫星微振动现象出现的主要因素,微振动现象严重阻碍了光学遥感卫星向更高分辨率方向发展。为了在设计时更有效地抑制光学遥感卫星微振动现象,需要较好地把握反作用飞轮微振动相关参数。本文提出了一种反作用飞轮连续转速状态下的参数辨识方法,即在时频域内获取飞轮扰振变化特性。首先利用时频分析获取非稳定干扰力信号的时频分布;然后利用加扇形窗的线性最小二乘估计方法辨识反作用飞轮的倍频特征;其后建立与飞轮转速相关的参数化模型,并利用加权非线性最小二乘估计方法辨识出参数化模型的待估参数,其后将模型参数转化为模态参数;再结合模态参数和倍频参数,获取飞轮干扰力或力矩的转速-频率峰值点。最后通过一组反作用飞轮测试试验验证本文所提出的方法,试验结果显示,连续转速状态下的辨识结果更接近反作用飞轮真实工作状态,该方法准确掌握了飞轮干扰力和力矩变化趋势,相对于传统固定转速状态下的测试结果平均误差小于1%,结果具有良好的精度。

Owing to manufacturing and assembling errors,the harmonic interference generated by a reac⁃tion wheel assembly(RWA)is the primary factor leading to the micro-vibration of optical remote sensing satellites,which seriously hinders the development of high-resolution optical remote sensing satellites.To achieve more effective micro-vibration suppression in the design of optical remote sensing satellites,it is necessary to accurately grasp the relevant parameters of RWA micro-vibrations.This paper proposes a parametric identification method for RWA at continuous rotating speeds.The micro-vibration characteris⁃tics of the RWA are obtained in the time-frequency domain.First,a time-frequency analysis is used to obtain a time-frequency distribution of non-stationary disturbing force/moment signals.Then,a linear least squares(LLS)estimator with a sector window is established to identify the harmonic characteristics.Sub⁃sequently,a speed-dependent parameterized model is built to describe the RWA model;further,using a weighted nonlinear least-square(WNLS)estimator,the parameters of the parameterized model are esti⁃mated and transformed into modal parameters.By combining modal parameters and harmonic characteris⁃tics,the speed-frequency peaks of the disturbing forces/moments can be estimated.Experiments are per⁃formed to verify the proposed method.The results demonstrate that the proposed method presents practi⁃cal trends of disturbing forces/moments.Compared with the results obtained under traditional fixed-speed states,the errors are less than 1%,confirming the accuracy of the proposed method.