摘要
为量化在役油气管道长期因地层运动而受到挤压的结构变形程度和避免油气泄漏等重大事故,基于电磁探伤测试仪MTD-J检测的脉冲涡流检测信号,提出一种基于脉冲涡流的井下油管形变估算方法(DEM),以及时发现管道形变严重的位置。在DEM中,对脉冲涡流响应信号进行预处理,去除外界因素引起的涡流信号基线漂移和油管连接处节箍信号。使用EMD方法对预处理后的信号进行分解,得到多个信号分量,再根据相关系数和均方根误差进行重构,从而去除混合噪声。最后采用基于改进粒子群算法的LSSVR模型对特征量进行训练,建立量化模型,实现油管形变的量化。实验结果表明:DEM的量化结果更符合实际情况,提高量化精度,降低平均误差和训练时间,具有较好的算法收敛性,比SVR和PSO-LSSVR更优。
In order to quantify the deformation degree of oil pipeline due to stratum movement and avoid major accidents such as oil and gas leakage,based on the pulse eddy current detection data detected by the electromagnetic detector MTD-J,an deformation estimation method(DEM)of downhole oil pipeline based on pulsed eddy current is proposed,and the location of severe pipeline deformation can be found in time.Firstly,DEM pre-processes the response signal of pulse eddy current to remove the data baseline drift caused by external factors and coupling signal of oil pipe joint.Then,it uses EMD method to decompose the preprocessed data to obtain multiple IMF,and reconstructs the signal according to the correlation coefficient and root mean square error to remove the mixed noise.Finally,it uses the LSSVR model based on the improved particle swarm optimization algorithm to train the characteristic variables,and establishes the quantitative model to realize the deformation quantification of oil pipeline.The experimental results show that the quantization results of DEM are more in line with the actual situation,then it improves the quantization accuracy and convergence,and reduces the average error and training time.DEM is better than SVR and PSOLSSVR.
作者
施佳椰
王章权
徐菲
刘半藤
陈友荣
周莹
SHI Jiaye;WANG Zhangquan;XU Fei;LIU Banteng;CHEN Yourong;ZHOU Ying(School of Information Science&Engineering,Changzhou University,Changzhou 213164,China;College of Information Science and Technology,Zhejiang Shuren University,Hangzhou 310015,China;Petroleum Engineering Institute,Zhongyuan Oilfield Company,SINOPEC,Puyang 457000,China)
出处
《中国测试》
北大核心
2021年第8期22-30,共9页
China Measurement & Test
基金
国家科技重大专项(2016ZX05017-003)
浙江省自然科学基金(LQ19F010012)。
