基于广义回归神经网络的管道泄漏精确定位方法

来源期刊:中南大学学报(自然科学版)2011年第z1期

论文作者:陈琦 林伟国

文章页码:943 - 948

关键词:管道泄漏;精确定位;移动窗相关分析;广义回归神经网络

Key words:pipeline leak; accurate locating; moving-window correlation analysis; generalized regression neural network (GRNN)

摘    要:针对复杂工况下管道泄漏定位不准问题,提出2种改进方法。提出移动窗相关定位算法,以一定步长在固定长度上下游信号中移动提取一定跨度信号做相关计算,取相关系数最大的位移作为泄漏信号的时间差,以克服干扰信号引起的定位不准问题。根据实测介质出站、入站的温度和流量,结合管道参数和定位机理模型,得到管道沿途的温度、声速、介质流速和理论时间差分布。以理论时间差为模型输入,以对应的管道各点位置为期望输出,实时建立基于广义回归神经网络的管道泄漏定位模型,结合实测信号时间差,实现泄漏点的精确定位。实际应用结果表明:该方法定位精度高,计算速度快,能够实时、可靠地解决泄漏定位问题。

Abstract: Two modified approaches for solving inaccurate problem of pipeline leak location in complex conditions were introduced. A locating approach based on moving-window correlation analysis was proposed, which extracted appropriate span signals from a fixed size upstream and downstream signals by adequate steps for correlation analysis and took the displacement with the maximal correlation coefficient as time difference of leak signals to overcome inaccurate locating problem caused by interference signals. According to the temperatures and flows of upstream and downstream which were measured online, combining with the parameters of pipeline and the model of locating mechanisms, the distribution of temperature, speed of sound, flow rate of transmission and time difference in theory along the pipeline were acquired. Taking the time difference in theory as input, and the corresponding locations of pipeline as output, the model based on generalized regression neural network (GRNN) real-timely was established. Combining with the time difference of leak signals acquired online, accurate leak location can be implemented. Actual application results show accuracy and effectiveness of the proposed approaches, which enables real-timely and reliably solve the leak locating problems.

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号