| With the continuous expansion of urban gas pipeline network coverage and the continuous updating of pipeline technology,urban gas pipelines have covered residential areas in most cities in China,bringing convenience to people's lives.However,due to the long-term burial and underground of urban gas pipelines,combined with corrosion,vibration,aging and other factors,it is easy to cause material leakage in the gas pipeline,which will cause serious personal injury and huge economic losses.At present,China's main gas pipeline state inspection method is mainly end-to-end manual instrument testing,and pipeline robot imaging exploration is supplemented.In view of the large number of gas pipeline networks,the complex structure of new and old pipeline networks,and the narrow space reserved inside and outside the pipeline,the end-to-end manual instrument testing method currently used in the mainstream cannot cover the aging detection of the gas official website.Therefore,the use of pipeline robots for pipeline network to detect faults in networked pipelines is an inevitable trend in the development of gas pipelines in China.Based on the commercial pipeline robot,CCTV(Closed-Circuit Television)and laser imaging system are used to collect the functional defects and structural defects of the gas pipeline respectively.An automatic defect recognition based on machine vision is proposed.Algorithm to improve the accuracy and efficiency of automated pipeline inspection.The main research contents of this paper are as follows:(1)A local enhancement method for CCTV image of gas pipeline based on median filtering method is proposed.Aiming at the characteristics of small inner diameter of gas pipeline,large amount of infiltration and uneven distribution of gas residue,this paper proposes a gas pipeline CCTV image processing noise reduction algorithm.The algorithm introduces the concept of gray-light intensity disturbance factor.Firstly,qualitative analysis is carried out on the light intensity information in the pipeline to confirm the respective characteristics of the gray-light intensity disturbance factor.Secondly,quantitative correlation of light intensity information is performed for the gray-light intensity disturbance factor.The different gray threshold compensation amounts are divided into sections,and after the binarization processing,only the contour extraction results of the disturbance light intensity qualified value and the unqualified value are output;finally,the secondary screening isperformed globally,and the final fault detection map is output.(2)A pattern recognition mechanism for pipeline functional defects based on BP fuzzy neural network is proposed.Due to the many characteristic parameters of gas pipeline defects,the method of manual research or “one-size-fits-off” pattern recognition in the system is not conducive to comprehensive evaluation of pipeline defects.Therefore,this paper combines the characteristics of gas pipeline defects,confirms the three functional defects of circularity,eccentricity and compactness,and uses it as the feature vector of pattern recognition.In addition,BP fuzzy neural network is introduced to identify defect patterns.The results of defect identification comparison between the damaged gas pipeline and the non-destructive gas pipeline in the laboratory environment are given.(3)Combined with 3D laser imaging technology for structural identification of gas pipelinesIn view of the lack of depth information on the panoramic view of the inner wall of the pipe outputted by the CCTV detection technology,it is impossible to effectively identify structural defects such as pipe deformation,foreign matter infiltration,and deposits.This paper introduces a three-dimensional laser imaging system to supplement it,quickly extract the three-dimensional structure of the pipeline,and make up for the shortage of high false alarm rate of CCTV detection.It has strong pertinence for structural defect detection. |
PDF Full Text Request