基于球形视频机器人的管道内缺陷检测方法研究

Research on Inner Surface Defect Detection Method of Pressure Pipe Based on Spherical Video Tube Defect Detection Robot

传统的压力管道内表面缺陷检测管道运行安全十分重要,目前检测方法对待检测管道的形状结构要求较高,检测方法低效耗时费力。设计了一种球形视频管道内缺陷检测机器人,通过搭载的高分辨率立体相机拍摄视管道内部频流,编制基于深度卷积神经网络软件对视频流进行带表面缺陷图片分选,采用Fast-RCNN快速区域卷积神经网络对管道缺陷进行标记,通过该标记可实现管道内部缺陷的安全评估,实验结果表明了该检测方法的有效性。

The traditional inner surface defect detection device is difficult to be used in the pipeline with complicated shapes, and the detection method is inefficient and time-consuming. A spherical video tube defect detection robot is designed. A software based on deep convolution neural network was developed to detect and classify the inner surface defects of the pipeline by capturing the video streams using a high resolution stereo camera. The Fast-RCNN fast region convolution neural network model was used to mark pipeline defects. The experimental results show the effectiveness of the proposed method.