基于经验模态分解和BP神经网络的油气两相流流型辨识

Flow pattern identification of oil-gas two-phase flow based on empirical mode decomposition and BP neural network

基于经验模态分解(empirical mode decomposition,EMD)和BP神经网络,提出了油气两相流流型辨识的新方法。应用EMD将差压信号分解成不同频率尺度上的单组分之和,并提取组分的归一化能量作为流型辨识特征量。BP神经网络以这些能量特征量为输入对油气两相流不同流型(包括泡状流、塞状流、层状流、弹状流和环状流)进行分类。实验结果表明,本文提出的流型辨识方法是有效的,其中泡状流、塞状流、层状流、弹状流和环状流的辨识精度分别为100%、89.4%、93.3%、96.3%和96.9%。

Based on empirical mode decomposition （EMD） and back propagation （BP） neural network, a new method is proposed to identify flow pattern of oil-gas two-phase flow. EMD is applied to the differential pressure signal of two-phase flow to obtain frequency components with different scales, and the normalized energy of the component is extracted as the feature of flow pattern identification. With these energy features as inputs, five flow patterns such as bubble flow, plug flow, stratified flow, slug flow and annular flow are identified using BP neural network. The experimental results indicate that the proposed BP-based method is effective for the identification of flow patterns ; and the identification rates are 100% , 89.4% , 93.3% , 96.3% , and 96.9% for bubble flow, plug flow, stratified flow, slug flow and annular flow respectively.