序贯概率比检验法在导航精度试飞中的应用
2014-07-18徐武军侯玉宏段亚
徐武军 侯玉宏 段亚
摘 要: 以往导航精度试飞采用基于参数估计理论的误差统计法,须完成规定的试飞架次才能得出试飞结论。这里根据假设检验理论总结出序贯概率比检验法,在给定置信度和误差概率时,求出导航精度的拒绝域、接收域和观察域;当试验曲线收敛于接收域时即可提前结束试飞,达到节省架次的目的。通过应用表明该方法不但适用于导航系统,对其他系统也具参考价值。
关键词: 序贯概率比检验; 导航精度; 飞行试验; 置信度; 误差概率
中图分类号: TN96?34 文献标识码: A 文章编号: 1004?373X(2014)11?0124?05
Abstract: The error statistical method based on theory of parameter estimation used in navigation precision flight test can not gain test conclusion until the scheduled flight testing sortie are completed. The sequential probability ratio test method is proposed in this paper according to the testing hypothesis theory, by which refused region, received region and observed region of navigation precision can be deduced while the confidence level and error probability are stipulated. The test may be permitted to be closed ahead of schedule only when test curve converges to the received region, and the flight test sortie is saved then. The method not only is applicable to the navigation systems but also has a reference value to other systems.
Keywords: sequential probability ratio test; navigation precision; flight test; confidence level; error probability
0 引 言
飞行试验是一项复杂的系统工程,耗资大,周期长。如何利用更少的试飞架次,客观准确地评估被试对象的功能和性能是试飞工程师始终面临的问题。导航精度评估的传统方法是误差统计法,其原理是认为导航误差服从正态分布,按照参数估计理论求出试飞样本量,计算出系统误差和随机误差,与研制总要求规定的指标相比较,得出导航精度是否满足要求[1]。本文根据Wald在四十年代发展起来的序贯概率比检验法[2],当给定置信度和误差概率时求出导航精度的拒绝域、接收域和观察域,然后分析导航精度是否满足指标;此方法的关键是随时观察误差试验的过程,当试验曲线收敛于接收域时就可提前结束试飞;与传统的误差统计法相比往往节省试飞架次,提高了试飞效率。
1 序贯概率比检验法原理[3?5]
1.1 导航精度试飞架次计算
由[m]次试……
