基于希尔伯特—黄变换与Elman神经网络的气液两相流流型识别方法

Applied Study of Hilbert-huang Transform and Elman Neural Network on Flow Regime Identification for Gas-liquid Two-phase Flow

气液两相流的流型对其流动和传热特性有很大的影响,所以如何确定流型一直是两相流研究中的重要课题。但是,由于两相流介质之间存在着随机多变的相界面,致使两相流的流型不仅是多种多样,而且其变化也带有随机性,这给流型识别带来了很大困难。而希尔伯特-黄变换(HHT)和神经网络在气液两相流流型识别中还很少见,文中提出了希尔伯特-黄变换与Elman神经网络相结合的气液两相流流型识别的新方法。将压差波动信号经验模态分解(EMD)后的固有模态函数(IMF)进行分析,提取IMF能量特征作为Elman神经网络的输入特征向量,对水平管内的气液两相流流型进行识别。实验结果表明:该方法能很好地识别水平管内的4种流型,为流型识别开辟了一条新的途径;另外,该方法优于BP网络且稳定、识别率高,具有可行性。

Flow and heat transfer characteristic of gas-liquid two-phase flow are strongly affected by its flow pattern. Therefore, the study on flow pattern is always an important subject of two-phase flow. However, as existing multifarious interphase boundary among the medium of two-phase flow, it leads to various of two-phase flow pattern, and the changes are random. So it is difficult to identify the flow pattern. And it is seldom to apply study of Hilbert-Huang transform （HHT） and neural network on flow regime identification for gas-liquid two-phase flow. In this article a flow regime identification method using Hilbert-Huang transformation combined with Elman neural network is put forward. Firstly the method analyzes the intrinsic mode function （IMF） obtained after the empirical mode decomposition （EMD）, then extracts IMF energy feature as the input feature vectors of the Elman neural network, lastly flow regime identification of the gas-liquid two-phase flow in a horizontal pipe can be performed. The experimental result shows that this method can identify the four flow regimes of gas-liquid two-phase flow in horizontal pipe. This method develops a new direction for the flow regime identification. In addition, the experimental result shows that this method is superior to BP neural network, and it is stable and higher identification. Result also proves that the method is feasible.