摘要
随着我国经济的发展与能源需求的增长,集输管线在天然气运输过程中得到了广泛应用。针对集输管线流体状态识别问题,采用超声兰姆波的非侵入式测量技术,提出一种基于弹性网络回归的集输管线流体状态辨识方法。以超声兰姆波测量信号为基础提取时频特征,挖掘信号中的全面信息;采用弹性网络回归进行特征筛选,保留其中的关键特征;利用筛选后的时频特征,构建随机森林模型。进行在线识别时,提取并保留实时数据的关键时频特征,输入至训练好的随机森林模型中,开展流体状态分类与预警。试验结果证明,所提方法在兰姆波检测信号下具有有效性。
With the development of Chinese economy and the increase in energy demand,gathering and transportation pipelines have been widely used in natural gas transportation.To address the problem of fluid state identification in gathering and transportation pipelines,a fluid state identification method for gathering and transportation pipelines based on elastic net regression is proposed by using the non-invasive measurement technology of ultrasonic Lamb wave.Based on the ultrasonic Lamb wave measurement signal,its time-frequency features are first extracted to mine the comprehensive information in the signal.Then the key features are selected using elastic net regression.The selected time-frequency features are then used to construct a random forest model.For online identification,the key timefrequency features in real-time data are extracted and retained,and the trained random forest model is used to realize the classification and early warning of the fluid state.The experimental results demonstrate that the proposed method is effective for Lamb wave detection signals.
作者
王俊峰
张宝雷
刘金海
王璐瑶
尚尊轩
WANG Junfeng;ZHANG Baolei;LIU Jinhai;WANG Luyao;SHANG Zunxuan(CNOOC Energy Technology&Services-Oil Production Co.,Tianjin 300452,China)
出处
《天津科技》
2025年第8期19-23,31,共6页
Tianjin Science & Technology
基金
海油发展采油科技项目“集输管线流通状态在线检测评估与治理技术研究及应用”(CYKJ-2023-09)。
关键词
集输管线
超声兰姆波
时频特征
弹性网络回归
流体状态辨识
gathering and transportation pipelines
ultrasonic Lamb wave
time-frequency feature
elastic net regression
fluid state identification
