管道天然气多组分浓度在线监测方法

A Real-Time Monitoring Method of Components and Concentrations of Pipeline Natural Gas

为了解决管道内天然气监测技术中由于各组分互相交叉而带来的组分与浓度识别精度低且无法在线监测的难题,提出了一种采用气体传感器阵列与信息融合技术相结合的多组分浓度在线监测方法。该方法首先通过传感器阵列获得对混合气体的交叉干扰响应值,然后利用比较结果最优的辨识回归算法MLP对测量的实验样本进行组分辨识后对交叉干扰量进行修正并融合,得到相关权重与偏置参数,最终经过非线性计算获得满足精度要求的浓度值。仿真结果表明:采用该方法对开源数据库中6种单一组分进行种类识别,整体分类准确性最高达到99%,对浓度进行回归识别中,回归系数R 2最高达0.99。将该方法用于管道内天然气混合气体的种类与浓度识别,可得到甲烷、非甲烷总烃、氮气的回归相关系数R 2分别为0.99、0.96和0.99。实验结果表明,该方法降低了交叉敏感的影响,满足了设备所需的精度要求,可以实现天然气的在线监测。

An on-line monitoring method of components and concentrations of pipeline natural gas is proposed to solve the problems that the recognition precision of components and concentrations in pipeline natural gas is low and the chromatography cannot be used online due to the cross interference of gas sensor array.This method combines a gas sensor array and information fusion technology.At first,the cross-interference response values of mixed gases are obtained through the sensor array,and the existing identification and regression algorithm MLP is used to identify the component gases and to correct the cross-interference quantity.Then the correlation weights and offset parameters are obtained.Finally,the concentration values satisfying precision requirements are obtained through nonlinear calculation.Simulation results of identifying six individual components in the open source database show that the overall classification accuracy of the proposed method is up to 99%,and the regression coefficient R 2 is up to 0.99 in the regression identification of the concentrations.When the method is applied to the identification of the components and concentrations of natural gas in a pipeline,the values of R 2 of methane,non-methane hydrocarbon and nitrogen are 0.99,0.96 and 0.99,respectively.These results show that the method reduces the influence of cross sensitivity,meets the accuracy requirements of equipment,and realizes on-line monitoring of natural gases.