基于XGBoost-GRA-DEMATEL面向任务携行航材消耗预测方法

A Mission-Oriented Aircraft Spare Parts Carried Consumption Prediction Method Based on XGBoost-GRA-DEMATEL

为提高执行任务时航材携行数量保障的科学性,充分考虑任务中各类影响因素,采取XGBoost算法对携行航材需求进行预测。首先,分析不同任务中影响航材消耗的各种因素,按照全面系统、科学简明等原则建立预测特征体系;其次,采用GRA,XGBoost,DEMATEL对特征重要性和相关度进行定性和定量分析并筛选,建立精简版特征集合;再次,用网格搜索法调参,提高模型预测的准确率和运行效率;最后,通过算例分析,并与GBDT,SVM算法对比分析,验证该方法在样本数据有限、影响因素多的情况下,可降低预测误差,避免过拟合,有较好的实用性和高效性。

In order to improve the scientificity of the quantity guarantee of aircraft spare parts carried during the mission and fully consider all kinds of influencing factors in the mission,the XGBoost algorithm is adopted to predict the demand of aircraft spare parts carried.Firstly,various factors affecting aircraft spare parts consumption in different missions are analyzed,and a predictive feature system is established according to the principles of comprehensiveness,systematization,science and conciseness.Secondly,GRA,XGBoost,DEMATEL algorithm are used to analyze and screen the importance and relevance of features,and a simplified version of feature system is established.Thirdly,the grid search method is used to adjust parameters to improve the accuracy and efficiency of model prediction.Finally,through example analysis and comparative analysis with GBDT,SVM algorithms,it is verified that this method can reduce the prediction error and avoid over fitting in the case of limited sample data and many influencing factors,and has good practicability and efficiency.