摘要
自行搭建了带提升管的内循环流化床试验台,研究了提升管风速、气化室风速、颗粒平均粒径、床层高度对循环流率的影响。基于遗传算法优化BP神经网络原理,建立了GA-BP人工神经网络模型,用来预测带提升管的内循环流化床的颗粒循环流率。通过对GA-BP神经网络模型颗粒循环流率的预测值与试验值的比较发现:当隐含层数目为22时,最大相对误差为±6.6917%,误差的均方差为2.899%。该模型预测数据与试验值比较吻合,能够较好的预测颗粒循环流率。
It was tested and analysed how gasification air velocity,draft tube velocity,particle average size and static bed height affect particle circulating flow rate on the test-bed of internally circulating fluidized bed with draft tube built by students.Based on BP neural network principle for genetic algorithm optimization,a GA-BP artificial neural network model was established to predict particle circulating flow rate.By comparing particle solids circulating flow rates got by GA-BP neural network model prediction with those got from experiment, the results showed that:when Neuron number of hidden layer is 22,the maximum relative error ranges between-6.6917% and 6.6917% and mean-square deviation of the error is 2.899%.The model predicted data coincide with the test values well,so the model can excellently predict particle circulating flow rates.
出处
《锅炉技术》
北大核心
2013年第1期26-31,共6页
Boiler Technology
基金
国家自然科学基金项目(50876030)
关键词
遗传算法
BP神经网络
颗粒循环流率
内循环流化床
genetic algorithm
BP neural network
particle circulating flow rate
internally circulating fluidized bed