【摘要】 管流截面流型作为描述气固两相流的重要参数之一 ,极大地影响着两相流动压力损失和传热传质等特性 ,同时还影响着其他参数 (如流量、分相含率等 )的准确测量以及流动系统的运行特性。传统的检测方法由于难以获得能真正反映流型的管道截面局部分布的实时信息 ,在工业应用中受到了限制。有鉴于此 ,在光学层析成像技术的基础上 ,提出了一种基于学习矢量量化神经网络的气固两相流流型识别方法 ,详细介绍了这种网络的结构、学习算法、训练样本集的确定等。通过计算机仿真 ,实验结果表明此方法对于气固两相流的 8种流型能有较好的识别能力 ,为两相流参数检测提供了一种新的思路与方法更多还原

【Abstract】 The flow regime,as one of significant parameters characterizing the gas-solid two-phase flow,not only affects strongly its performances on pressure loss,heat transfer,mass transfer and so on,but also affects the measurement accuracy of other parameters (such as flux,phase holdup,etc.)as well as the operating performance of flowing system.The application of traditional measurement methods is limited in industry because they have some difficulties in really reflecting real-time and local distribution information of flow regime in the section of pipe.As a result,based on optical tomography,presented a method to recognize the gas-solid two-phase flow regime by means of LQV neural network whose structure,learning algorithm,training sample collections and so on were described in detail.The simulating results show that this method can preferably recognize the eight flow regimes of gas-solid two-phase flow.So a novel method is provided to measure the parameters of two-phase flow.更多还原