针对航空产品工程变更不易预测传播路径的问题,本文利用信息论和统计学的有关原理,提出了一套基于经验的工程变更传播路径预测方法。利用熵和互信息的概念,从海量经验数据中获取在预测过程中起到重要作用的参数;以其为样本空间,提出了一种从语义和变更影响两个角度计算变更参数相似度的算法;基于贝叶斯公式,分析了多因素情况下,相关参数的被影响概率,以此作为判断变更路径的依据;最后,以某型号航空涡轮叶盘为例对本文方法进行了验证,结果表明本文方法具有一定的准确度和优越性。
In view of the difficulties of engineering change propagation in aerospace products, a new history-based propagation path predicting method is proposed. Important parameters are obtained using the concepts of entropy and mutual information. Then, a parameter-similarity algorithm is proposed from the aspect of semantics and change-effect. Furthermore, the regulation and method of change propagation predicting are analyzed based on Bayes rules. Finally, an experiment of managing an aerospace turbine disk is performed to validate our method and it shows its validity and merits.