应用神经网络测量凝析天然气流量

为了精确测量凝析天然气的气液相流量,研究了将神经网络用于气液两相流流量测量的方法。以中国石油大学自动化系试制的流量检测仪表样机为硬件基础,分析了仪表的差压波动信号,并通过比较F比值确定了特征提取方法。根据多相流存在多种流型的特点,采用了首先识别流体流型再计算其流量的方法。研究表明:神经网络在凝析天然气流量测量中具有较大的应用潜力。但要获得较高的层量精度,还需要更为准确的信号特征提取方法以及更加合理的网络结构。

Use Neural Network to Measure Wet Gas Flow

Neural network are used for wet gas flow measurement with high precision.Differential pressure signals from the flowmeter prototype developed by the China University of Petroleum are analyzed. Different techniques for feature extraction are studied and their F ratios compared.Considering the characteristic of wet gas that has different flow patterns,a process that identifies flow pattern first and then calculates flow rate is adopted. The research shows that: neural network can be used in the wet gas flow measurement with satisfactory results.The gas phase error reached 10 % . Accurate ways for feature extraction and rational network structure both play an important role in the measurement .