CAJ | 학술논문

论述了车载轨道巡检系统结构及各子系统设计方案。巡检系统由轨道图像采集、数据分析和数据管理3个子系统构成。图像采集子系统采用线阵 CCD 等间距运动扫描获取轨道图像。基于机器学习理论构建了数据分析子系统，可对钢轨表面伤损、扣件异常进行智能识别。数据管理子系统可对检出缺陷进行组合查询并输出报表。该系统目前已应用于我国各高速铁路干线，对轨道主体设备外观进行普查，最高检测速度可达160 km/h。
논술료차재궤도순검계통결구급각자계통설계방안。순검계통유궤도도상채집、수거분석화수거관리3개자계통구성。도상채집자계통채용선진 CCD 등간거운동소묘획취궤도도상。기우궤기학습이론구건료수거분석자계통，가대강궤표면상손、구건이상진행지능식별。수거관리자계통가대검출결함진행조합사순병수출보표。해계통목전이응용우아국각고속철로간선，대궤도주체설비외관진행보사，최고검측속도가체160 km/h。
T his paper discussed the vehicle-mounted track inspection system structure and its sub system design scheme. T he inspection system is composed of 3 sub system including track image acquisition,data analysis and data management. T he image acquisition system obtained the track image by equidistance motion scanning with linear array CCD. Based on the machine learning theory,the data analysis system is built for intelligent recognition of the rail surface damage and abnormal fastener. Data management system is built for querying fault record and generating the inspection report. T he inspection system has been applied to routine appearance inspection for the main track equipment in many high-speed railways,the highest detection speed of which can arrive at 160 km/h.