基于改进型BP网络的气液两相流流型识别  

Flow Patterns Identification of Air-water Two-phase Based on Improved BP Neural Network

作　　者：杨萍[1] 杨良煜[2] 张玉杰[1] 

机构地区：[1]陕西科技大学电信学院,陕西西安710021 [2]西安科技大学电控学院,陕西西安710054

出　　处：《计算机测量与控制》2009年第9期1828-1830,共3页Computer Measurement &Control

基　　金：陕西省工业攻关项目(2006K05-G18)

摘　　要：根据气液两相流的特点,将其分为层状流、塞状流、弹状流和环状流等流型,分别对其采集差压信号,并利用概率密度函数(PDF)对差压信号特征进行了分析,得出了PDF的4个特征参数,即波峰个数K_1、波峰峰值K_2、波峰位置K_3、PDF方差K_4。将K_1,K_2,K_3,K_4构成的特征向量作为神经网络的输入样本对BP神经网络进行训练并进行流型识别。结果表明,该方法具有准确率高等优点,是流型识别的一种新手段。Air--water two--phase flow can be divided into stratified flow, churn flow, slug flow and annular flow according to its flow patterns. Differential pressure signal of the two--phase flow is measured. The probability density function （PDF） is used to extract the characters of the differential pressure signal. Four characteristic parameters of the PDF, the peak number K1, the maximum peak value K2, the peak position K3, and the PDF variance K4 are defined. The characteristic vectors consisted of the four defined parameters are used to train the BP neural network in order to identify the flow regimes. Results show this method has accurate identification. So it is a new method for flow patterns identification.

关 键 词：流型识别 BP网络 概率密度函数 

分 类 号：TQ021.1[化学工程]