Chanan Gary, Nelson Jerry
University of California, Irvine, California 92697, USA.
Appl Opt. 2009 Nov 10;48(32):6281-9. doi: 10.1364/AO.48.006281.
The active control systems of segmented mirror telescopes are vulnerable to a malfunction of a few (or even one) of their segment edge sensors, the effects of which can propagate through the entire system and seriously compromise the overall telescope image quality. Since there are thousands of such sensors in the extremely large telescopes now under development, it is essential to develop fast and efficient algorithms that can identify bad sensors so that they can be removed from the control loop. Such algorithms are nontrivial; for example, a simple residual-to-the-fit test will often fail to identify a bad sensor. We propose an algorithm that can reliably identify a single bad sensor and we extend it to the more difficult case of multiple bad sensors. Somewhat surprisingly, the identification of a fixed number of bad sensors does not necessarily become more difficult as the telescope becomes larger and the number of sensors in the control system increases.
分段镜望远镜的主动控制系统容易受到少数（甚至一个）分段边缘传感器故障的影响，其影响可能会在整个系统中传播，并严重损害望远镜的整体图像质量。由于目前正在研制的极大望远镜中有数千个这样的传感器，因此开发快速高效的算法来识别故障传感器，以便将其从控制回路中移除至关重要。这样的算法并不简单；例如，简单的拟合残差测试往往无法识别故障传感器。我们提出了一种能够可靠识别单个故障传感器的算法，并将其扩展到更具挑战性的多个故障传感器的情况。有点令人惊讶的是，随着望远镜变大以及控制系统中传感器数量的增加，识别固定数量的故障传感器并不一定会变得更加困难。
