基于RBF的油气水段塞流流型超声识别方法  

Ultrasound recognition method for flow patterns in oil-gas-water slug flow based on RBF neural network

作　　者：苏茜 夏志飞 刘振兴[1,2] SU Qian;XIA Zhifei;LIU Zhenxing(School of Information Science and Engineering/School of Artificial Intelligence,Wuhan University of Science and Technology,Wuhan 430081,Hubei,China;Engineering Research Center of Metallurgical Automation and Measurement Technology,Wuhan University of Science and Technology,Wuhan 430081,Hubei,China)

机构地区：[1]武汉科技大学信息科学与工程学院/人工智能学院,湖北武汉430081 [2]武汉科技大学冶金自动化与检测技术教育部工程研究中心,湖北武汉430081

出　　处：《化工进展》2024年第2期628-636,共9页Chemical Industry and Engineering Progress

基　　金：国家自然科学基金(61903281,61901423,51907144);湖北省自然科学基金(2019CFB145);中国博士后科学基金(2018M642932);武汉市知识创新专项曙光计划(2022010801020311)。

摘　　要：石油管道内的多相流流型识别主要集中在气液两相流和油水两相流方向且准确识别流型范围有限,为了解决油气水段塞流流型识别问题,本文提出一种基于RBF神经网络和超声传播规律的油气水段塞流流型识别方法。根据油气水段塞流的相分布特点,建立了流型识别超声测试仿真模型。采用超声透射衰减技术和反射回波技术研究水平管道油气水三相段塞流超声响应特性,提取透射衰减信号区分段塞流液膜区、气泡夹带区和稳定液塞区。利用反射信号时间序列数据中的回波能量等统计特征,通过RBF神经网络对油气水段塞流进行流型识别。结果表明,基于超声传播机理以及RBF神经网络三相段塞流流型识别率为95.7%。基于RBF神经网络的流型识别算法研究为超声技术实现水平管油气水段塞流流型识别提供了理论基础。Flow pattern recognition plays a crucial role in the efficient operation and management of oil pipelines.However,the existing methods primarily focus on gas-liquid and oil-water two-phase flow,with limited accuracy in identifying flow patterns within the oil-gas-water slug flow segment.To address this limitation,this study proposed an ultrasound-based method for identifying flow patterns in the oilgas-water slug flow segment using a radial basis function(RBF)neural network.The proposed method utilized the unique characteristics of phase distribution within the oil-gas-water slug flow segment and establishes a comprehensive set of 350 ultrasound test simulation models.By employing ultrasound transmission attenuation and reflection echo techniques,the response characteristics of the oil-gas-water slug flow segment within the pipeline were investigated.The transmission attenuation signals were then extracted to differentiate between the liquid film region,bubble entrainment region,and stable liquid slug region.To classify the flow patterns,the statistical features,such as the energy of reflected signal time series data,were extracted and utilized as inputs for the RBF neural network.The experimental results demonstrated that the proposed method achieves a high flow pattern recognition rate of 95.7%based on the ultrasound propagation mechanism and RBF neural network.This research provided a theoretical foundation for implementing flow pattern recognition of oil-gas-water slug flow in horizontal pipelines using ultrasound technology.The application of the RBF neural network-based recognition algorithm significantly enhanced the accuracy and efficiency of flow pattern identification,offering valuable insights for the effective operation and control of oil pipeline systems.

关 键 词：多相流 瞬态响应 油气水段塞流 超声衰减 RBF网络 流型识别 

分 类 号：TE81[石油与天然气工程—油气储运工程] TB551[理学—物理]