论文部分内容阅读
为提高航班运行风险预测精确度,参照民航局咨询通告《航空承运人运行控制风险管控系统实施指南》,首先分析山东航空航班控制工作流程,初步筛选出15个航班运行风险评估指标项;然后精选100个航班历史数据,根据粗糙集理论,结合遗传算法和Johnson算法约简评估项,获取8个核心指标;最后,利用支持向量机(SVM)算法建立风险预测模型,并用Matlab进行仿真。结果表明:对于高中低3类风险等级,用该方法所得样本分类正确率可达82.22%,该方法可用于航班运行风险的评估和分级。
In order to improve the accuracy of flight operation risk prediction, reference is made to CAAC Advisory Circular “Implementation Guide for Air Carrier Operation Control Risk Management and Control System”. Firstly, Shandong Airlines flight control workflow is analyzed, and 15 flight risk assessment index items are initially screened. Select 100 flight history data, according to rough set theory, combined with genetic algorithm and Johnson algorithm to reduce the evaluation items, access to eight core indicators; Finally, the use of support vector machine (SVM) algorithm to establish a risk prediction model and simulate with Matlab. The results show that for the high, middle and low risk levels, the correct classification rate of the samples obtained by this method can reach 82.22%. This method can be used to evaluate and classify the flight operation risk.