A Computationally Efficient Algorithm for Disturbance Cancellation to Meet the Requirements for Optical Payloads in Satellites
A Computationally Efficient Algorithm for Disturbance Cancellation to Meet the Requirements for Optical Payloads in Satellites
复制标题
DOI：
--
复制
发表时间：
2001-09
期刊：
影响因子：
--
通讯作者：
Christian G. R. Taranti
中科院分区：
文献类型：
--
作者：
Christian G. R. Taranti
【摘要】：振动控制是卫星中一个非常重要的问题。新型高分辨率数字成像设备对振动特别敏感。激光通信中使用的天线还需要非常安静的环境，以免其性能下降。斯图尔特平台能够将光学有效载荷与嘈杂的航天器总线隔离。直到最近，所有隔振应用中都只使用被动方法。数字信号处理技术的最新进展使得振动控制算法的开发成为可能，但这些算法通常需要大量的计算能力。这项工作探索了通过使用六足机构对光学有效载荷使用一种计算高效的振动隔离方法。如果设备是线性的，该方法会抑制指定频率处的振动，并且不会影响未指定的频率。数学分析包括收敛分析和输出中未分配频率的影响。评估该算法的计算要求并与多重误差最小均方进行比较。该方法对非线性具有很强的鲁棒性；其性能可与多误差最小均方法相媲美，但计算时间和内存需求仅为其一小部分。它还需要很少的植物知识。通过使用单输入/单输出设备和非线性六足模型的模拟验证了理论结果。该控制器还在两个不同的六足机构中进行了实验验证，发现当使用有噪声的参考信号时，其性能与使用多误差最小均方法获得的性能相似或更好。
Abstract : Vibration control is a very important issue in satellites. The new high-resolution digital imaging devices are especially sensitive to vibrations. Antennas used in laser communications also require a very quiet environment so that their performance is not degraded. The Stewart platform is capable of isolating an optical payload from the noisy spacecraft bus. Until recently only passive methods were used in all vibration isolation applications. Recent advances in Digital Signal Processing techniques made the development of vibration control algorithms possible but these usually require large computational power. This work explores using a computationally efficient vibration-isolation method for optical payloads by using hexapods. The method suppresses the vibration at the assigned frequencies and does not affect unassigned frequencies if the plant is linear. The mathematical analysis includes convergence analysis and the effect of unassigned frequencies in the output. The computational requirements of the algorithm is evaluated and is compared to the Multiple-Error Least Mean Square. The method is very robust to nonlinearities; its performance is comparable to the Multiple-Error Least Mean Square with a fraction of the computational time and memory requirements. Also it requires very little plant knowledge. Theoretical results are verified through simulations using a Single-Input/Single-Output plant and a nonlinear hexapod model. The controller was also experimentally validated in two different hexapods and the performance was found to be similar to or better than the performance obtained with the Multiple-Error Least Mean Square method when a noisy reference signal is used.