基于神经网络的蜂窝陶瓷蓄热体温度效率预测  

Temperature efficiency prediction of regenerative honeycomb ceramics based on neural network

作　　者：彭振 陆金桂[1] Peng Zhen;Lu Jingui(School of Mechanical and Power Engineering,Nanjing Tech University,Jiangsu Nanjing,211816,China)

机构地区：[1]南京工业大学机械与动力工程学院,江苏南京211816

出　　处：《机械设计与制造工程》2024年第8期96-100,共5页Machine Design and Manufacturing Engineering

摘　　要：针对蜂窝陶瓷蓄热体在通风瓦斯蓄热氧化过程中温度效率难以直接预测的问题,提出一种基于神经网络的蜂窝陶瓷蓄热体温度效率预测方法。首先选择蜂窝陶瓷蓄热体单孔截面形状、截面积、壁厚、高度和高温烟气入口速度作为研究对象,其次采用计算流体动力软件模拟仿真通风瓦斯蓄热氧化过程,最后将仿真计算所得的230组蜂窝陶瓷蓄热体温度效率数据作为学习样本,30组数据作为验证样本,同时研究隐含层节点数与BP神经网络预测精度之间的关系。结果表明,当隐含节点数为12时,神经网络的最优平均预测误差为0.63%,可以用于快速准确预测通风瓦斯蓄热氧化过程中蜂窝陶瓷蓄热体的温度效率。Focusing on the difficult to predict directly the thermal efficiency of regenerative honeycomb ceramics in the process of regenerative oxidation of gas,the prediction method for thermal efficiency of regenerative honeycomb ceramics based on a neural network is presented.Firstly,five objects of the shape,the cross-sectional area,the wall thickness,the height of honeycomb and the flue gas velocity in regenerator of regenerative thermal oxidation equipment is selected to study,then the process of regenerative oxidation of gas is simulated by CFD software.Finally,230 sets data from numerical simulation is used as study sample,30 sets data from numerical simulation is used as validation sample.It investigates the relationship between the number of hidden layer nodes and the prediction accuracy of BP neural network.The result shows that when the number of hidden layer nodes is 12,optimization mean relative prediction deviation of neural network is 0.63%,which can be used to predict the temperature efficiency of regenerative honeycomb ceramics in process of regenerative oxidation of gas rapidly and accurately.

关 键 词：蓄热蜂窝陶瓷 通风瓦斯 温度效率 BP神经网络 预测 

分 类 号：TK16[动力工程及工程热物理—热能工程]