基于图像识别的液压同步提升系统漏油在线检测  

Online detection of oil leakage in hydraulic synchronous lifting system based on image recognition

作　　者：刘锡炀 卞永明[1] 陈启凡 周怡和 蒋哲[2] 刘广军[1] LIU Xiyang;BIAN Yongming;CHEN Qifan;ZHOU Yihe;JIANG Zhe;LIU Guangjun(School of Mechanical Engineering,Tongji University,Shanghai 201804,China;Shanghai Salvage Bureau,Ministry of Transport of the People’s Republic of China,Shanghai 200090,China)

机构地区：[1]同济大学机械与能源工程学院,上海201804 [2]交通运输部上海打捞局,上海200090

出　　处：《中国工程机械学报》2022年第3期257-262,共6页Chinese Journal of Construction Machinery

基　　金：国家重点研发计划资助项目(2018YFC0309700)。

摘　　要：针对液压同步提升施工中可能出现的油液泄露问题,结合数字化监控系统,研究基于图像识别的液压同步提升系统泵站接头漏油在线检测技术。首先采用基于运动目标的漏油可疑区域提取,研究Vibe算法,并引入动态抽样因子φ提高背景建模的抗干扰能力;然后提取可疑区域的形状、颜色、纹理特征,为使特征向量维数达到最佳,使用随机森林算法进行特征筛选;采用随机森林进行可疑区域的分类,将动态惯性权重ω引入粒子群算法,利用其对随机森林的参数进行优化。利用PC、USBCAN和液压试验台对漏油检测算法进行测试。结果表明:提出的漏油检测算法对泵站接头漏油有良好的检测效果。Aiming at the oil leakage problem that may occur in hydraulic synchronous lifting construction,and combined with digital monitoring system to study the online detection of oil leakage based on image recognition.Firstly study of the Vibe algorithm based on motion target for oil leakage suspicious region extraction,and the dynamic sampling factor φ is introduced to improve the anti-interference ability of background modelling;then the shape,color and texture features of the suspicious region are extracted and the RF algorithm is used for feature screening in order to optimize the feature vector dimension;the RF is used to classify the suspicious region.The dynamic inertia weights ω is introduced into the particle swarm optimization(PSO)algorithm,which is used to optimize the parameters of the RF.The oil leakage detection algorithm is tested using PC,USBCAN and hydraulic bench.The results prove that the oil leakage detection algorithm proposed in this paper have good detection effect on the oil leakage of pump station joints.

关 键 词：液压同步提升 远程监控 Vibe 随机森林 粒子群 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]