基于混合神经网络参数优化的两相流流型识别方法  

Flow pattern identification of two-phase flow based on an optimized hybrid neural network

作　　者：王萌[1] 张松 施艳艳[1] 杨珍 史水娥[1] Wang Meng;Zhang Song;Shi Yanyan;Yang Zhen;Shi Shuie(Henan Key Laboratory of Optoelectronic Sensing Integrated Application,Henan Normal University,Xinxiang 453007,China)

机构地区：[1]河南师范大学河南省光电传感集成应用重点实验室,河南新乡453007

出　　处：《河南师范大学学报(自然科学版)》2025年第3期121-127,共7页Journal of Henan Normal University(Natural Science Edition)

基　　金：国家自然科学基金(61903127);河南省高校科技创新人才支持计划(21HASTIT018).

摘　　要：针对气液两相流传感器测量数据的强非线性和非平稳性导致流型识别困难的问题,提出一种基于混合神经网络参数优化的流型识别方法.所提方法首先采用滑动窗口法将传感器测得的不同流型电导率数据分割为若干子序列,再利用变分模态分解算法获得各子序列的固有模态函数,通过提取固有模态函数的Hjorth特征数据集实现对各子序列非线性特征的描述.接着,将随机森林算法引入卷积神经网络的分类层,进而构建混合神经网络,并采用鲸鱼算法对混合神经网络中3个超参数进行优化.最后,采用优化后的混合神经网络对Hjorth参数特征向量数据集进行分类,进而实现流型识别.实验结果表明,所提方法对4种流型的平均辨识准确率达到98.52%.Accurate flow pattern identification is a great challenge due to strong nonlinearity and non-stationary of measurement data in gas-liquid two-phase flow.In this work,a new flow pattern identification method based on optimized hybrid neural network is proposed.In the proposed method,the measured conductivity data of different flow patterns is firstly segmented using the sliding window method.Then,the variational mode decomposition algorithm is utilized to obtain the intrinsic mode functions of each subsequence.The nonlinear features of each subsequence are described by extracting the Hjorth features from the intrinsic mode functions.Next,the random forest algorithm is introduced into the classification layer of the convolutional neural network to construct a hybrid neural network.The whale algorithm is employed to optimize the three hyperparameters in the hybrid neural network.Finally,flow pattern is identified by classifying the feature vector dataset of Hjorth parameters with the optimized hybrid neural network.The experimental results show that an average identification accuracy of 98.52% can be realized for the four flow patterns when the proposed method is used.

关 键 词：气液两相流 Hjorth参数 混合神经网络 随机森林 

分 类 号：O359[理学—流体力学]