灾害学2018，Vol.33Issue(2)：193-199,7.DOI:10.3969/j.issn.1000-811X.2018.02.034
基于机器学习的大雾天气背景下特强浓雾本地化诊断研究
Study on the Localization Diagnosis of Extra Heavy Fog on The Background of the Fog Weather Based on Machine Learning Algorithms
摘要
Abstract
Low visibility fog is an important cause of traffic accidents.Prediction of extra heavy fog is a difficult problem in weather forecast.Based on the Lianyungang 58044 station 2014-2016 hourly meteorological data,selection of fog occurs,the machine learning decision tree algorithm is lower than 50m of strong visibility fog establish diagnosis model based on Meteorological factors.The results show that the diagnosis model of CART decision tree algorithm can visually and accurately diagnose the extra heavy fog based on the model,and has high generalization ability,learning by accounting for about 75 ％ of the study sample data,the accuracy of the model of learning ability is 90.04％,the remaining 25％ sample data on the left and right model generalization test,testing accuracy is 82.25％;from the decision tree model can be found,compared with other grades of fog,ultra extra heavy fog is the most critical factor is the temperature;diagnostic effect of LSVM algorithm for extra heavy fog is best,but can understand the low degree of complexity,higher than the CART algorithm is easy to use.关键词
大雾/特强浓雾/机器学习/本地化/诊断模型/CART决策树算法Key words
Fog/Extra heavy fog/Machine learning/Localization/Diagnosis model/CART Algorithm分类
资源环境引用本文复制引用
史达伟,李超,史逸民,张银意..基于机器学习的大雾天气背景下特强浓雾本地化诊断研究[J].灾害学,2018,33(2):193-199,7.基金项目
淮河流域气象开放研究基金(HRM201602) （HRM201602）
江苏省气象局2014年现代化项目“江苏海洋气象综合业务平台” （）
江苏省科技厅项目(BE2011720) （BE2011720）
连云港市科技局项目(SH1422,SH1013) （SH1422,SH1013）
江苏省气象局青年基金(KQ201802) （KQ201802）
