摘要
本文探讨利用人工智能中的启发式搜索技巧确定观测值粗差的方法。除了利用由残差建立的统计检验量外,本文首次引入残差间的相关系数作为另一重要的启发信息,并启发式地使用不同大小的置信度来解决粗差被“掩盖”和正常观测值被“淹没”的问题。文中讨论了启发式搜索粗差的程序设计,并以解析相对定向为例进行了试验。试验结果表明,利用启发式搜索来探测粗差可以克服传统的粗差探测方法的缺点,具有应用前景。
:
In this paper the heuristic search techniques of artificial intelligence are used to -discover and locate gross errors. The correlation coefficients between two test values are introduced as another important heuristic knowledge besides the statistical test values, and different risk levels are heuristically used to overcome the 'swamping' effects of normal observations and the 'masking' effects of gross error. The programs of heuri stic search of gross' errors are written for analytical relative orientation and used to demostrate their applicability. The result of the experiment shows that the use of heuri stic search methods in discovering and locating gross errors can overcome the defaults of existing methods.
出处
《测绘学报》
EI
CSCD
北大核心
1990年第4期250-256,共7页
Acta Geodaetica et Cartographica Sinica