基于智能识别与周期检测的钢轨伤损自动预警方法研究

Research on Automatic Early Warning Method for Rail Flaw Based on Intelligent Identification and Periodic Detection

目前采用人工回放方式对钢轨探伤车的检测数据进行分析,并与同一线路周期检测的钢轨伤损记录表进行对比,形成当次检测报告,实现钢轨伤损预警。人工回放与对比分析过程效率较低。提出一种基于智能识别技术与周期检测的钢轨伤损自动预警方法,结合线路基础信息数据库,通过对钢轨探伤车检测数据中的超声波反射体进行智能识别,输出超声波反射体的类别与位置,生成含有各类超声波反射体的序列(区段检测数据),基于编辑距离法将区段检测数据与区段标准检测数据(由周期检测数据生成)进行自动对比,形成检测报告,提高了数据分析效率。通过对某铁路集团有限公司5个区段的检测数据进行伤损自动预警测试,将自动预警结果与人工预警结果进行对比,验证了该方法的有效性和高效性。

Currently, manual playback method is used to analyze the data generated by the rail flaw detection vehicle, which is compared with the rail flaw record of periodic detection. According to the comparison, cur rent detection report is formed to realize the early warning of the rail flaw. However, the manual playback and comparison analysis are inefficient. Based on intelligent recognition and periodic detection, an automatic early warning method for rail flaw was proposed. Combined with the railway line infrastructure database, through the intelligent recognition of ultrasonic reflectors in the inspection data of rail flaw detection vehicle, the cate gory and location of the ultrasonic reflectors were identified. Then, the sequences containing various types of ultrasonic reflectors（section detection data） were generated. Based on the Editing Distance method, the section detection data was automatically compared with standard detection data generated by periodic detection. After wards, the detection report was generated automatically and effectively instead of manual comparison analysis. Through comparison between automatic warning and artificial warning on five sections of China Railway Some Group Co., I.td., the results show the effectiveness and efficiency of this method.