摘要
针对由于异常路况数据量较小,导致数据集数据不均衡问题,从而引起路况预测模型准确度不高以及运行效率低等问题,提出一种基于SmoteEnn;GBoost的路况感知方法。设计研发了路况数据采集系统,捕获并处理实时的路况数据;使用SmoteEnn算法对数据集进行数据合成处理;采用XGBoost算法对提取的主要特征进行训练和测试,从而识别出正常路况和异常路况。结果表明,基于SmoteEnn;GBoost的路况感知方法,对比目前较常用的分类模型SVM、逻辑回归、GBDT、随机森林等,在提高路况分类效果的同时大幅缩短算法的运算时间。
Aiming at the problem of data imbalance in the data set due to the small amount of abnormal road conditions,resulting in low accuracy of the road condition prediction model and low operating efficiency,a road condition sensing method based on Smooth Enn_XGBoost is proposed.Designed and developed a road condition data collection system to capture and process real-time road condition data;use the Smooth Enn algorithm to perform data synthesis processing on the data set;use the XGBoost algorithm to train and test the extracted main features to identify normal road conditions and abnormal road conditions.The results show that the road condition perception method based on Smooth Enn_XGBoost,compared with the more commonly used classification models SVM,logistic regression,GBDT,random forest,etc.,can greatly reduce the calculation time of the algorithm while improving the effect of road condition classification.
作者
杨黎娜
姚凯学
何勇
席雷鹏
刘文才
赵继露
YANG Lina;YAO Kaixue;HE Yong;XI Leipeng;LIU Wencai;ZHAO Jilu(School of Computer Science and Technology,Guizhou University,Guiyang 550025,China)
出处
《智能计算机与应用》
2021年第11期137-142,147,共7页
Intelligent Computer and Applications
