Abstract:

In order to improve the attitude detection effect of a micro-satellite, aiming at the panoramic fish-eye imaging characteristics of the static infrared Earth sensor, the imaging model is established with a single infrared image as the object of study. An algorithm combining the interval vector product with the confidence interval mean of t location-scale distribution is proposed. The infrared imaging is projected onto the space sphere to form a spatial image ring; the vector product is made by taking the points at intervals on the image point ring, and the sample set of the attitude angle of the current target is obtained; the sample mean in the confidence interval of 1σ is calculated again, and the estimation of the attitude angle of the target is finally worked out. Compared with the result of the original algorithm, the standard deviation of the target attitude angle, within ±10 degree, is reduced from 0.06 degrees to 0.03 degrees, and the accuracy is improved by 50%. In addition, in order to avoid the time-consuming problem of iterative convergence, a table checking method used in practical engineering is given, and the operation time is reduced by 87%. The measurement accuracy and efficiency of the static infrared Earth sensor are improved.

Key words: Static infrared horizon sensor, Algorithm optimization, Vector product, t location-scale distribution, Confidence interval, Micro-satellite