摘要
飞行事故率是表征飞行安全水平的重要指标,其预测是典型的小样本问题.针对目前飞行事故率预测中存在的预测精度不高的问题,提出了一种基于回归支持向量机的飞行事故率预测建模方法.最后结合实际算例,采用SVR进行了飞行事故率预测建模并把预测结果与灰色预测和灰色马尔柯夫链预测进行了对比.仿真结果表明SVR具有很高的建模精度和泛化能力,从而验证了采用SVR进行航空飞行事故率预测的合理性和先进性.
Flight accident rate is an important index which reflects the aviation safety degree. The prediction of flight accident rate is typically a small sample problem. Regarding the problem of low prediction precision in today＇s prediction of accident rate, this paper brought forward Support Vector Regression （SVR） and applied it into the prediction of flight accident rate. Finally, according to the example, the model of flight accident rate prediction based on SVR was built and the results were compared with grey model and grey-Makov prediction model. The simulation results show that SVR has high modelling precision and strong generalization. Therefore the method brought forward above is valid and advanced.
出处
《数学的实践与认识》
CSCD
北大核心
2009年第8期124-128,共5页
Mathematics in Practice and Theory
基金
国家自然科学基金(60572172)
关键词
支持向量机
回归支持向量机
飞行事故
事故率预测
Support Vector Machine （SVM）
SVR
flight accident
prediction of accident rate
