基于经验模式分解和概率神经网络的气液两相流识别.pdfVIP

27 卷 17 期 中 国 电 机 工 程 学 报 Vol.27 No.17 Jun. 2007 2007 年6 月 Proceedings of the CSEE 2007 Chin.Soc.for Elec.Eng. 文章编号：0258-8013 (2007) 17-0072-06 中图分类号：O359 文献标识码：A 学科分类号：47010 基于经验模式分解和概率神经 网络的气液两相流识别 斌，周云龙，向新星，窦华荣 东北电力大学能源与机械工程学院，吉林省 吉林市 132012 ） Identification Method of Gas-liquid Two-Phase Flow Regime Based on Empirical Mode Decomposition and Probabilistic Neural Network SUN Bin，ZHOU Yun-long，XIANG Xin-xing，DOU Hua-rong （School of Energy Resources and Mechanical Engineering , Northeast Dianli University, Jilin 132012, Jilin Province, China ） ABSTRACT：Aiming at the non-stationary characteristics of (intrinsic mode function，IMF)之和，再选取若干个包含主要 differential pressure fluctuation signals of gas-liquid two-phase 流型信息的 IMF 分量进行进一步分析。由于流型转变时， flow, and back propagation neural networks (BPNN) like slow 压差波动信号各频带的能量会发生变化，因而可以从各IMF convergence of learning and liability of dropping into local 分量中提取能量特征参数作为神经网络的输入参数来识别 minima, flow regime identification method based on empirical 流型。对水平管内空气水两相流4 种典型流型的识别结果 mode decomposition (EMD) and probabilistic neural network is 表明，EMD 能量比小波包能量特征具有更高的流型识别率， put forward. First of all, original signals are decomposed into a 可以准确、有效地识别流型。 finite number of stationary Intrinsic Mode Functions (IMF), and 关键词：热能动力工程；气液两相流；流型识别；经验模式 then a number of IMF containing main flow regime information 分解；概率神经网络 is selected for the further analysis. The energy of acceleration 0 引言 differential pressure fluctuation signal in diffe