基于电容层析法糖浆气液两相流识别  

Identifying of gas-liquid two-phase flow regime for sugar syrup based on electrical capacitance tomography

作　　者：刘梦伦[1] 傅文利[1] 蒙华[2] 刘金花[1] 赵进创[1] 

机构地区：[1]广西大学计算机与电子信息学院,广西南宁530004 [2]广西医科大学信息中心,广西南宁530000

出　　处：《广西大学学报（自然科学版）》2011年第5期792-795,共4页Journal of Guangxi University（Natural Science Edition）

基　　金：国家自然科学基金资助项目(60762001);广西高等学校优秀人才计划(桂教人才0804);广西大学科研基金资助项目(XJZ110647)

摘　　要：针对目前常规两相流流型识别方法在糖厂煮糖罐糖浆识别中精度低的缺点,提出基于电容层析法(ECT)的气液两相流流型判别软测量模型。首先对由ECT传感器采集的检测数据优化处理并提取特征参数作为BP神经网络的辅助变量,气液两相流相应实际流型作为BP网络的主导变量,通过神经网络结构参数的优化训练,构建出二级并行的自组织竞争神经网络模型,进而实现两相流流型识别。仿真实验结果表明,该糖浆气液两相流流型识别模型精度较高、判别速度快,有较好的工业应用前景。Aiming at the low accuracy of general flow pattern recognition for boiling sugar syrup in the sugar processing industry,a vapor-liquid two-phase flow pattern identification soft measuring model based on electrical capacitance tomography is proposed.Firstly,detected data of ECT sensor were optimized and then the imputing characteristic parameters were extracted and used as the auxiliary variable of BP neural network,and the corresponding flow pattern worked as the BP network main parameters.Then two parallel self-organization competitive neural networks were constructed.Finally a real-time flow pattern recognition model was achieved after neural network being trained.The simulation results show that this method has the advantages of low cost and high accuracy and has the potential in the industry.

关 键 词：ECT 两相流 识别 BP神经网络. 

分 类 号：TP216[自动化与计算机技术—检测技术与自动化装置]