基于小波变换与神经网络的上倾管流型识别方法

为实现上倾管气液两相流流型的智能识别,提出了基于小波变换与概率神经网络的流型识别方法。采用中国石油大学(华东)室内小型环道试验装置进行气液两相流试验,采集上倾管流型以及相应的持液率信号。运用小波变换对持液率信号进行5级分解,并对分解后的信号提取标准差作为概率神经网络的输入参数,对试验中获得的分层流、气泡流、段塞流、严重段塞流流型进行识别。结果表明:该方法对4种流型的识别效果较好,其整体识别率为96.5%,其中分层流和严重段塞流的识别率高达98%。基于小波变换与概率神经网络的上倾管流型识别方法能够有效克服传统识别方法中主观因素的影响,不仅显著提高了流型识别的准确率,而且识别过程更加智能。(图5,表3,参22)

Identification method for flow pattern in upward pipe based on wavelet transform and neural network

In order to realize the intelligent identification of gas-liquid two-phase flow patterns in upward pipes,the flow pattern identification method based on wavelet transform and probabilistic neural network was proposed.The gas-liquid two-phase flow test was conducted using the small indoor loop experimental device of China University of Petroleum(East China),and the flow pattern in upward pipes and the corresponding liquid holdup signals were acquired.Wavelet transform was used to decompose the liquid holdup signal into five levels,and the standard deviation of the decomposed signal was extracted as the input parameter of probabilistic neural network to identify the stratified flow,bubble flow,slug flow and severe slug flow patterns obtained in the experiment.The results show a good recognition effect of the method on four flow patterns.The overall recognition rate is 96.5%,and the recognition rate for stratified flow and severe slug flow can be up to 98%.The identification method for flow pattern in upward pipe based on wavelet transform and probabilistic neural network can effectively overcome the influence of subjective factors in traditional recognition methods,significantly improve the accuracy of flow pattern recognition,and also facilitate a more intelligent recognition process.