基于槽式孔板与神经网络的湿气流量计  

Wet Gas Flowmeter Based on Slotted Orifice and Neural Network

作　　者：石岗[1] 耿艳峰[1] 

机构地区：[1]中国石油大学信息与控制工程学院,山东东营257061

出　　处：《自动化仪表》2009年第5期46-48,共3页Process Automation Instrumentation

基　　金：国家自然科学基金资助项目(编号:60672003)

摘　　要：在研究某湿气流量计样机的基础上,提出了一种应用槽式孔板与神经网络技术实现湿气流量计量的方法。简要介绍了槽式孔板的特点及流量计样机的结构,采用统计分析与相关分析相结合的方法对信号特征量进行筛选,并应用神经网络技术对数据处理、构建二级神经网络系统。此设计实现了流型识别和计量,利用网络集成技术进一步提高网络系统的计量精度和泛化能力。现场测试结果表明,应用该流量计进行湿气流量计量,其气相累积流量计量误差为3%,液相累积流量计量误差为6%,满足了生产计量的精度要求。On the basis of the prototype of certain wet gas flowmeter, the metering technology for wet gas flow based on slotted orifice and neural network （ NN ） is proposed. The features of the slotted orifice and the structure of the prototype are introduced briefly. By adopting the method combining statistical analysis and correlation analysis, the features of measuring signals are screening selected, the data are processed by using neural network technology. The double level NN system is structured to implement flow type identification and flow metering respectively; and with network integration technology, the metering accuracy and generalizing capability are enhanced. The result of field test shows that the cumulated error of gas phase is 3% ; and the cumulated error of liquid phase is 6% ;which meets the requirement of production.

关 键 词：槽式孔板 神经网络 网络集成 特征提取 湿气 计量技术 

分 类 号：TP216.1[自动化与计算机技术—检测技术与自动化装置] TE372[自动化与计算机技术—控制科学与工程]