一种基于半监督学习的危险品运输车辆运行远程监控方法

A Remote Monitoring Method of Dangerous Goods TransportVehicle Operation Based on Semi-Supervised Learning

为提高车辆运行时的定位精度,提出基于半监督学习的危险品运输车辆运行远程监控方法。将危险品运输车辆的行驶路线采用二维栅格进行划分,采用分类回归模型降低运输车辆行驶路线的计算误差,计算危险品运输车辆行驶路线的预测值,增加标记分类的概率置信度,利用二维卷积核离散运算实时更新车辆远程定位数据。并通过实例对比分析基于半监督学习的车辆运行远程监控、北斗定位、5G通信技术、OBD系统等方法的测试精度。实例测试结果表明:在采样点数量为200时,基于半监督学习的车辆运行远程监控方法相较于其他3种方法,行驶路程均方误差最小,监测到的行驶距离最接近实际测量结果,实时定位精度最高。

In order to improve the positioning accuracy of vehicles,a remote monitoring method for dangerous goods transportation vehicles based on semi-supervised learning is studied.Initialization of dangerous goods transport vehicle route occupancy grid,based on semi-supervised learning algorithm,calculation of vehicle remote distance,real-time update of vehicle remote location data.The example test results show that in the mean square error calculation where the number of sampling points is 200,the mean square error of this method is 0.264,with high real-time positioning accuracy and excellent real-time positioning effect,which provides reference for ensuring the safety of dangerous goods transport vehicles and timely reaching the destination.