摘要
微波相移法原油含水率传感器检测是实现高含水原油含水率在线检测的有效手段之一,但其检测精度易受矿化度的影响。针对高含水原油中大量存在的矿化度组份(NaCl和CaCl_2),试验研究了不同比例及含量的双组份矿化度对微波相移法原油含水率检测传感器精度的影响,得出了双组份矿化度(NaCl和CaCl_2)对原油含水率检测精度的影响规律。由于矿化度的组份及含量与原油含水率检测值的关系受多种因素的影响,很难建立准确的误差补偿模型。为此,建立误差校正的BP神经网络模型,该模型把微波相移法原油含水率传感器的检测误差从13. 912%降低到1. 821%,提高了检测精度。数据对比结果表明:BP神经网络预测模型优于多元线性回归预测模型。
Microwave moisture sensor for crude oil is one of the effective means to realize on-line detection of water content in high water cut crude oil,but its detection accuracy is easily affected by salinity. A large number of salinity fractions( NaCl and CaCl2) exist in water bearing crude oils,the influence of the salinity of the two components with different proportions and contents on the moisture content of the crude oil by microwave method was studied,the influence law of two component salinity( NaCl and CaCl2) on the detection accuracy of crude oil moisture content by microwave method was obtained. It is difficult to establish an accurate error compensation model due to the influence of various factors on the relationship between the components and content of salinity and the detection value of water content. To solve the above problem,the Back-Propagation( BP) neural network model of error correction is established and the model can reduces the detection error of microwave moisture content sensor from13. 912 % to 1. 821 %,which improves the detection accuracy. The data comparison shows that BP neural network prediction model is better than multivariate linear regression prediction model.
作者
董鹏敏
江接波
吕沛志
李晓辉
肖艳鹏
Dong Pengmin;Jiang Jiebo;Lii Peizhi;Li Xiaohui;Xiao Yanpeng(Department of Mechanical Engineering, Xi' an Shiyou University, Xi’an 710065 , China)
出处
《现代制造工程》
CSCD
北大核心
2019年第3期96-101,共6页
Modern Manufacturing Engineering
基金
陕西省自然科学基础研究计划-重大基础研究项目(2016ZDJC-11)
西安石油大学硕士研究生创新基金项目(2015cx140430)
关键词
原油含水率
双组份矿化度
微波传感器
检测误差
BP神经网络
误差校正
water content of crude oil
two component of salinity
microwave moisture content detection sensor
detection error
BP neural network
error correction
