基于小波和Elman神经网络的气液两相流流型识别方法

传统的流型识别方法仅可作为一种定性的流型识别方法。为了克服传统流型方法的不足,采用小波分析和El-man神经网络技术来实现气液两相流流型的智能识别,测量了水平管内气液两相流的压差波动信号,应用小波分析对流型的动态压差波动信号进行分析、提取特征,然后将小波能量作为Elman神经网络的输入,从而实现对流型的智能识别。实验结果证明,该方法能够很准确地识别出4种流型,并且具有很好的识别效果,从而为流型的在线识别提供了一种定量的流型识别方法。

A Method for Discriminating Gas-liquid Two Phase Flow Patterns Based on Wavelets and Elman Neural Networks

The traditional flow pattern discrimination method can only be used as a kind of qualitative flow pattern discrimination.To overcome such a defect of the traditional method,a wavelet analysis and Elman neural network technology has been adopted to realize an intelligent discrimination of gas-liquid two-phase flow patterns.In this connection,the pressure-difference fluctuation signals of the gas-liquid two-phase flow inside horizontal tubes were measured and a wavelet analytic method was employed to analyze the dynamic pressure-difference fluctuation signals of the flow patterns,extracting relevant characteristics.Then,the wavelet energy was treated as an input to the Elman neural network,thus accomplishing an intelligent discrimination of the flow patterns.The test results show that the method under discussion can identify 4 kinds of flow patterns with a very high accuracy.The method has a very good discrimination effect,thereby providing a quantitative flow pattern discrimination method for an on-line discrimination of flow patterns.