【作者】 刘洋；

【导师】 王振军；

【作者基本信息】 中国矿业大学 ， 交通运输（专业学位）， 2022， 硕士

【摘要】 在我国,铁路运输装载量大、稳定性高,是煤炭运输的主要方式。但是在寒冷的北方冬季,装载煤炭的列车长时间处在低温运行状态,车厢内含有水分的煤炭在运输途中就会和车厢内壁之间产生冻结现象,在达到目的地进行卸车作业时出现卸载不完全的情况,部分煤炭仍残留在车厢内。目前检测车厢冻煤存量的方法主要还是人工观测,由工人沿着爬梯至车厢顶端进行目测,该方法不仅增加劳动力工作强度,精确度低,而且工作效率低,观测工人自身的安全性也难以保证。3D激光雷达技术由于其适用性强、效率高等优点逐渐成为研究的热点,广泛应用在非接触下的目标检测。三维激光雷达扫描技术可以实现扫描目标表面信息全覆盖,通过对扫描生成的点云数据解析,还原扫描物体表面真实特征。基于三维激光雷达的车厢冻煤存量检测技术能够克服传统检测技术的缺点,节省人工成本,提高检测效率。基于3D激光雷达的车厢冻煤存量检测,需要经过数据采集、数据预处理、点云切片、点云体积计算等流程,每一个流程对点云数据的处理都会影响到车厢冻煤存量的计算结果,因而在处理数据时要统筹兼顾、精准把握。论文研究设计了激光雷达扫描获取点云信息的硬件设备,激光雷达固定在可伸缩支架上,实现对车厢全方位精准扫描;在软件方面,利用VS 2019软件配置PCL库实现点云数据处理,具体内容有:(1)在点云数据预处理方面,针对扫描获取的点云数据,根据雷达和车厢的相对位置关系,提取车厢范围内的点云数据,对点云进行轮廓倾斜校正和去除运动畸变;然后对点云数据进行坐标系转换,利用移动位移融合算法拼接单帧点云;采用统计滤波和体素化网格法对提取的点云数据进行过滤和精简;同时,采用移动最小二乘法对点云进行平滑处理,消除点云“重影”问题。(2)针对点云数据切片,采用相等间距对车厢进行分割,将分割的点云投影后滤波,利用射线360度算法和alpha算法提取点云轮廓。(3)针对点云体积计算,利用Shoelace定理计算点云轮廓截面面积,与分割间距相乘,从而获得车厢点云的体积大小。论文设计了激光雷达检测车厢冻煤存量的算法,实现点云数据的高效处理,验证了相关算法的可行性;同时,该系统对使用激光雷达检测车厢其它物品的体积具有一定的参考意义和借鉴价值。更多还原

【Abstract】 In China,railway transportation has large loading capacity and high stability.It is the main way of coal transportation.However,in the cold northern winter,the train loaded with coal operates at low temperature for a long time,and the coal containing water in the carriage will freeze between the carriage and the inner wall of the carriage during transportation.When the unloading operation reaches the destination,there is incomplete unloading,and some coal is still left in the carriage.At present,the method of detecting the frozen coal stock in the carriage is mainly manual observation,which is carried out by workers along the ladder to the top of the carriage.This method not only increases the labor intensity,has low accuracy,but also has low work efficiency,and it is difficult to ensure the safety of observation workers.3D lidar technology has gradually become a research hotspot because of its strong applicability and high efficiency.It is widely used in non-contact target detection.Three dimensional lidar scanning technology can realize full coverage of scanning target surface information,and restore the real characteristics of the scanned object surface by analyzing the point cloud data generated by scanning.The detection technology of car frozen coal stock based on three-dimensional lidar can overcome the shortcomings of traditional detection technology,save labor cost and improve detection efficiency.The pre-processing process of cloud slicing and cloud slicing data of each car will affect the accurate processing of the data of the car’s stock,so the pre-processing of cloud slicing and cloud slicing data will be taken into account.This paper studies and designs the hardware equipment of laser radar scanning to obtain point cloud information.The laser radar is fixed on the telescopic support to realize the omni-directional accurate scanning of the carriage;In terms of software,use vs 2019 software to configure PCL library to realize point cloud data processing.The specific contents are as follows:(1)in terms of point cloud data preprocessing,for the point cloud data obtained by scanning,extract the point cloud data within the carriage according to the relative position relationship between radar and carriage,correct the contour inclination of point cloud and remove motion distortion;Then the coordinate system of the point cloud data is transformed,and the moving displacement fusion algorithm is used to splice the single frame point cloud;Statistical filtering and voxel grid method are used to filter and simplify the extracted point cloud data;At the same time,the moving least square method is used to smooth the point cloud to eliminate the "ghosting" problem of the point cloud.(2)For the point cloud data slice,the car is segmented with equal spacing,the segmented point cloud is projected and filtered,and the point cloud contour is extracted by ray 360 degree algorithm and alpha algorithm.(3)For the volume calculation of point cloud,the cross-sectional area of point cloud contour is calculated by using shoelace theorem,which is multiplied by the segmentation spacing,so as to obtain the volume size of car point cloud.This paper designs the algorithm of lidar to detect the frozen coal stock in the carriage,realizes the efficient processing of point cloud data,and verifies the feasibility of the relevant algorithm;At the same time,the system has certain reference significance and reference value for using lidar to detect the volume of other items in the carriage.更多还原