首页|期刊导航|哈尔滨商业大学学报(自然科学版)|一种管道清洁装置牵引性能及管道缺陷分析

一种管道清洁装置牵引性能及管道缺陷分析OA

Traction performance of pipe cleaning device and analysis of pipe defects

中文摘要英文摘要

针对城市污水管道堵塞及人工清理的局限性,开发了一种新型自适应变径履带式管道清洗检测机器人.该机器人周向布置三组履带驱动机构,前端搭载清扫刷与钻头复合机构,两侧配高压喷枪,内部集成摆杆螺母副变径机构与红外传感器.力学计算与有限元分析表明,机器人在900~1 200 mm 直弯管内自适应能力优异,最大牵引力与正压力线性正相关,关键连杆强度满足设计要求.同时,引入 YOLOv8 深度学习算法进行缺陷识别,实验表明在交并比 IoU=0.5 时,渗漏、变形、障碍物和结垢四类缺陷的平均精度均值mAP 达0.765,大幅提升了管道清淤与检测的智能化水平,为排水管网全生命周期运维提供了高效解决方案.

Aiming at the blockages of urban sewage pipelines and the limitations of manual cleaning,a novel adaptive variable-diameter tracked pipeline cleaning and inspection robot was developed.The robot was equipped with three sets of tracked drive mechanisms circumferentially,a compound mechanism consisting of a sweeping brush and a driller at the front end,and high-pressure spray guns on both sides,while a pendulum-screw variable-diameter mechanism and an infrared sensor were integrated inside.Mechanical calculations and finite element analysis demonstrated that the robot possessed excellent adaptability within straight and curved pipelines ranging from 900 mm to 1200 mm,where the maximum traction force was linearly and positively correlated with the normal pressure,and the strength of key links satisfied the design requirements.Meanwhile,the YOLOv8 deep learning algorithm was introduced for defect recognition.Experimental results showed that when the intersection over union(IoU)was 0.5,the mean average precision(mAP)for four types of defects—leakage,deformation,obstacle,and scaling—reached 0.765.This study significantly enhanced the intelligence level of pipeline clearing and inspection,providing an efficient solution for the lifecycle operation and maintenance of drainage pipe networks.

蒋东升;张文杰;方天阳;顾彬;金发阳

安徽建筑大学 机械与电气工程学院,安徽 合肥 230009安徽建筑大学 机械与电气工程学院,安徽 合肥 230009安徽建筑大学 机械与电气工程学院,安徽 合肥 230009安徽建筑大学 机械与电气工程学院,安徽 合肥 230009安徽建筑大学 机械与电气工程学院,安徽 合肥 230009

信息技术与安全科学

管道清洗装置牵引力分析履带式机器人管道缺陷分析变径机构深度学习

pipe cleaning devicetraction analysistracked robotpipe defect detectionvariable-diameter mechanismdeep learning

《哈尔滨商业大学学报(自然科学版)》 2026 (4)

461-468,512,9

国家基金青年项目(No.52405169)项目名称:考虑多样性接触特征耦合作用的颗粒类流态形成机理与润滑性能调控

评论