基于Faster R-CNN的铁路货车心盘螺栓智能检修

Intelligent maintenance of rail truck tray bolts based on Faster R-CNN

为了改善人工检修心盘螺栓浪费大量人力,检修作业效率过低的缺陷,研究基于Faster R-CNN的铁路货车心盘螺栓智能检修方法。利用机器视觉设备采集铁路货车心盘螺栓图像,对所采集图像进行膨胀与腐蚀处理,获取图像的感兴趣区域。设置图像的感兴趣区域作为Faster R-CNN模型的输入,Faster R-CNN模型利用RPN部分,生成螺栓状态检测区域,通过池化层与全连接层的运算,输出铁路货车心盘螺栓状态检测结果。针对非正常状态的铁路货车心盘螺栓,利用全自动拧紧机通过自动卡紧、自动对位拧紧等操作,完成心盘螺栓的全自动维修。实验结果表明,该方法可以精准检测铁路货车心盘螺栓的正常、松动、丢失等状态,全自动维修非正常状态的心盘螺栓,提升铁路货车心盘螺栓检修的作业效率。

In order to improve the defects of manual maintenance of the tray bolt,which wastes a lot of manpower and has low maintenance operation efficiency,the intelligent maintenance method of railway truck tray bolt based on Faster R-CNN is studied.Using machine vision equipment to collect the image of railway freight car core plate bolt,the acquired image is processed by expansion and corrosion,and the interested area of the image is obtained.The region of interest of the image is set as the input of the Faster R-CNN model.The Faster R-CNN model uses the RPN part to generate the bolt status detection region.Through the calculation of the pooling layer and the full connection layer,the bolt status detection results of the railway freight car core plate are output.Aiming at the abnormal state of railway freight car centerplate bolt,automatic tightening machine is used to complete the automatic maintenance of centerplate bolt by automatic clamping and automatic alignment tightening.The experimental results show that the method can accurately detect the normal,loose and missing state of the center bolt of railway freight car,automatically repair the abnormal state of the center bolt,and improve the working efficiency of the overhaul of the center bolt of railway car.