ZHANGJinping,NIRui,WANGYao,et al.Quantitative Research on Functional Defects in Drainage Pipelines Based on Deep Learning[J].China Water & Wastewater,2026,42(11):128-136.
基于深度学习的排水管道功能性缺陷量化研究
- Title:
- Quantitative Research on Functional Defects in Drainage Pipelines Based on Deep Learning
- Keywords:
- drainage pipeline; defect detection; instance segmentation; YOLO11; Edge-SAM; quantitative assessment
- 摘要:
- 针对排水管道功能性缺陷图像中存在的文字、箭头等标注信息干扰及量化精度不足等问题，单一模型难以在检测与分割间兼顾精度与鲁棒性。为此提出一种融合改进YOLO11检测框与点提示从而增强Edge-SAM的缺陷分割能力与提高缺陷量化精度的方法。首先，通过改进YOLO11骨干网络并引入自适应卷积与边界优化策略，实现轻量化条件下的高精度目标定位，将mAP@50提升至0.826；其次，采用聚类分析设计自适应提示融合策略，驱动Edge-SAM实现良好的像素级分割，mIoU达0.681 6；随后，基于圆截面几何约束与加权最小二乘拟合提取管道轮廓，结合缺陷面积比例计算实现精确量化；最后，通过参数化修正公式抑制数值偏离区间，使沉积、结垢、障碍物等缺陷的等级匹配准确率分别达0.68、0.73和0.77。该方法在检测精度、分割质量、模型部署和量化可靠性方面具有良好效果，为城市排水管网智能化运维提供了高效可行的技术方案。
- Abstract:
- To address the interference caused by annotation information such as text and arrows, as well as the insufficient quantification accuracy in images of functional defects in drainage pipelines, a single model often fails to balance detection precision and segmentation robustness. Therefore, this study proposed a method that integrated an improved YOLO11 detection framework with point prompts to enhance the defect segmentation capability of Edge-SAM and improve defect quantification accuracy. First, by optimizing the YOLO11 backbone network and introducing adaptive convolution and boundary refinement strategies, the method achieved high-precision object localization under lightweight conditions, increasing the mAP@50 to 0.826. Second, a clustering-based adaptive prompt fusion strategy was designed to guide Edge-SAM in achieving superior pixel?level segmentation, reaching an mIoU of 0.681 6. Subsequently, based on circular cross-sectional geometric constraints and weighted least squares fitting, the pipeline contour was extracted, and precise quantification was realized by calculating the defect area ratio. Finally, a parameterized correction formula was employed to suppress numerical deviation, resulting in accurate defect grade matching rates of 0.68, 0.73, and 0.77 for sedimentation, scaling, and obstacles, respectively. The method proposed achieves excellent performance in detection accuracy, segmentation quality, model deployment, and quantification reliability, providing an efficient and feasible technical solution for the intelligent operation and maintenance of urban drainage networks.
相似文献/References:
[1]李明.高水压条件下排水管道局部树脂固化修复及预处理[J].中国给水排水,2020,36(20):45.
LI Ming.Local Resin Curing Repair and Pretreatment Technology of Drainage Pipes under High Water Pressure[J].China Water & Wastewater,2020,36(11):45.
[2]周杨军,蒋仕兰,解铭,等.非开挖修复技术在城市排水管道维护中的应用[J].中国给水排水,2020,36(20):58.
ZHOU Yang-jun,JIANG Shi-lan,XIE Ming,et al.Application of Trenchless Repair Technology in Urban Drainage Pipeline Maintenance[J].China Water & Wastewater,2020,36(11):58.
[3]安关峰,张蓉,张欣,等.《城镇排水管道非开挖修复工程施工及验收规程》解析[J].中国给水排水,2020,36(20):71.
AN Guan-feng,ZHANG Rong,ZHANG Xin,et al.Analysis on The Specification for Construction and Acceptance of Trenchless Rehabilitation Engineering of Urban Drainage Pipeline[J].China Water & Wastewater,2020,36(11):71.
[4]曹井国,石东优,董泽樟,等.翻转式原位固化（CIPP）技术用于城市排水管道修复[J].中国给水排水,2021,37(6):128.
CAO Jing-guo,SHI Dong-you,DONG Ze-zhang,et al.Application of Rotary Cured-in-Place Pipe(CIPP) in Urban Drainage Pipe Rehabilitation[J].China Water & Wastewater,2021,37(11):128.
[5]刘琳,刘勇,黄宁君.新型原位热塑成型管道非开挖修复技术应用案例[J].中国给水排水,2021,37(6):134.
LIU Lin,LIU Yong,HUANG Ning-jun.Application Case of New Trenchless Pipe Rehabilitation Technology for Formed-in-Place Pipe (FIPP)[J].China Water & Wastewater,2021,37(11):134.
[6]桑稳姣,郝世文,陈志毅,等.基于鲁棒优化模型的排水管道疏浚污泥量测算[J].中国给水排水,2021,37(19):106.
SANG Wen-jiao,HAO Shi-wen,CHEN Zhi-yi,et al.Estimation of Dredging Sludge Volume in Drainage Pipe Based on Robust Optimization Model[J].China Water & Wastewater,2021,37(11):106.
[7]杜预,郭帅,潘刚,等.国内外排水管道状况评估系统对比分析研究[J].中国给水排水,2021,37(20):9.
DU Yu,GUO Shuai,PAN Gang,et al.Comparative Analysis of Pipelines Condition Assessment Protocols at Home and Abroad[J].China Water & Wastewater,2021,37(11):9.
[8]王健,刘国华,齐鲁,等.城市排水管道沉积物与污水间物质转移转化研究进展[J].中国给水排水,2021,37(24):34.
WANG Jian,LIU Guo-hua,QI Lu,et al.Research Progress in Material Transformation between Sediment and Sewage in Urban Drainage Pipeline[J].China Water & Wastewater,2021,37(11):34.
[9]袁艺鸣,张广毅,方宏远,等.排水管道微生物群落对管道腐蚀的影响[J].中国给水排水,2022,38(8):46.
YUANYi-ming,ZHANGGuang-yi,FANGHong-yuan,et al.Influence of Microbial Community in Drainage Pipeline on Corrosion[J].China Water & Wastewater,2022,38(11):46.
[10]田琪,叶建州,闻雪,等.《城镇排水管道原位固化修复用 内衬软管》团标解读[J].中国给水排水,2022,38(8):108.
TIANQi,YE Jian-zhou,WENXue,et al.Interpretation of China Urban Water Association Group Standard Municipal Sewer Pipelines Rehabilitation Using Cured-in-place Pipe—Lining Tubes （T/CUWA 60052-2021）[J].China Water & Wastewater,2022,38(11):108.
[11]凌同华,贝政豪,张胜,等.基于YOLO-Pipe和ByteTrack的排水管道缺陷检测[J].中国给水排水,2025,41(3):125.
LINGTong-hua,BEIZheng-hao,ZHANGSheng,et al.Surface Defect Detection in Sewer Pipelines Based on YOLO-Pipe and ByteTrack Methods[J].China Water & Wastewater,2025,41(11):125.