中心提升管内循环流化床颗粒循环流率试验与BP神经网络预测研究  

EXPERIMENT OF SOLID CIRCULATION RATE IN INTERNALLY CIRCULATION FLUIDIZED BED WITH DRAFT TUBE AND RESEARCH OF BP NEURAL NETWORK PREDICTION

作　　者：史洋[1] 尹萍 陈鸿伟[3] 郝青哲[1] 李崇[1] 

机构地区：[1]国网河北省电力公司电力科学研究院,石家庄050021 [2]河北华电石家庄鹿华热电有限公司,石家庄050200 [3]电站设备状态监测与控制教育部重点实验室(华北电力大学),保定071003

出　　处：《太阳能学报》2016年第6期1516-1520,共5页Acta Energiae Solaris Sinica

基　　金：国家自然科学基金(50876030);高校博士点基金(20090036110008)

摘　　要：自行设计并搭建中心提升管内循环流化床冷态试验台,就提升管风速、鼓泡床风速、鼓泡床静床高、床料平均粒径几方面因素对颗粒循环流率的影响进行系统的试验研究。试验结果表明:对于给定的床料,颗粒循环流率随两床风速的增大而增大;固定两床风速,颗粒循环流率随鼓泡床静床高的增大而增大,随物料平均粒径的增大而减小。利用Matlab神经网络工具箱,建立3层BP神经网络颗粒循环流率预测模型。预测结果表明:在隐含层神经元数量为6时,误诊率最小,预测相对误差在±9%以内,网络性能最优,能较好地预测颗粒循环流率。A cold test station of internally circulating fluidized bed with draft tube was designed and built. The effect of several factors such as gas velocities of the draft tube and bubbling fluidized bed, static bed height and the average size of particles in the beds on solid circulating rate was systematically studied. The experimental results indicate that the solid circulation rate increases with increasing gas velocities of the draft tube or bubbling fluidized bed. The solid circulation rate grows with increasing static bed height while declines with increasing particle size under constant gas velocities in the two beds. A three layers back propagation （BP） neural network model is built to predict the solids circulation rate based on Matlab neural network toolbox. The predicted results indicate that the misdiagnosis rate gets to the minimum with the relative error within ＋9% and the network property is in the optimum condition when the nodes of hidden layer are 6. The solid circulation rate can be well predicted with the model.

关 键 词：内循环流化床 提升管 颗粒循环流率 预测 BP神经网络 

分 类 号：TK513.5[动力工程及工程热物理—热能工程]