摘要
以粒子蜂群算法整合神经网络,提出一套可以预测高性能混凝土强度模型的方法论.以两个已经发表的方法进行比较,包括演化运算树及倒传递网络.由模型准确度可知,研究提出的三种不同隐藏层节点的粒子蜂群神经网络模型预测准确度高于演化运算树,但接近倒传递网络.由参数的影响性可知,粒子蜂群神经网络认为水泥、龄期、水、高炉矿渣粉、超塑剂、粉煤灰添加量对于高性能混凝土强度的影响性大,而粗、细骨料用量对高性能混凝土强度并不敏感,这样的结果与实际相符合.
This study used particle bee algorithm（ PBA） combined with artificial neural network（ NN） to predict the strength model of high- performance concrete（ HPC）. This study also compared the accuracy of the results with two proposed methods： genetic operation tree（ GOT） and back- propagation network（ BPN）. The results showed that three different hidden nodes design of particle bee neural network（ PBNN） are more accurate than GOT and closer to BPN. Besides,the impact of the amount of the parameters such as cement,age,water,blast- furnace slag,super plasticizer and fly ash aggregate have a large influence on the strength of HPC,while the amount of coarse and fine aggregate have a small influence on the strength of HPC. The influence analysis results are consistent with the practice.
出处
《福州大学学报（自然科学版）》
CAS
北大核心
2016年第2期253-258,共6页
Journal of Fuzhou University(Natural Science Edition)
基金
国家自然科学基金资助项目(51308120)
福建省自然科学基金资助项目(2014J05055)
福建省教育厅科技资助项目(GYZ15120)
关键词
粒子蜂群算法
高性能混凝土
演化运算树
倒传递网络
粒子蜂群神经网络
particle bee algorithm
high-performance concrete
genetic operation tree
back-prop agation network
particle bee neural network
