RBF神经网络在线学习方式及其在大坝位移监测中的应用  

Online study method of RBF neural network and its application in the dam displacement monitoring

作　　者：王德文 李智录[2] 卢瑞章[2] 

机构地区：[1]陕西省河务水库管理站,陕西西安710004 [2]西安理工大学水利水电学院,陕西西安710048

出　　处：《西北农林科技大学学报（自然科学版）》2006年第12期226-230,共5页Journal of Northwest A&F University(Natural Science Edition)

摘　　要：针对目前RBF神经网络训练算法存在的问题,提出了一种模拟人类学习方式的自动调整隐层节点数的在线训练方法,对其理论依据进行了分析,并用实例对其进行了验证。结果表明,此种学习方法速度快、拟合精度高、新旧知识均可记忆,克服了以往算法的不足,具有很大的实用性。Aiming at the present problem in RBF neural network,an on-line training algorithm is presented,which can simulate the study of human and modify the nodes automatically. The theory on which this algorithm is based has been discussed. Results show that the algorithm can study fast and memorize both new and old knowledge with high simulating precision. So the algorithm can overcome the shortcomings of other existing algorithms and has great practicability.

关 键 词：RBF神经网络 在线训练 大坝位移监测 径向基函数 

分 类 号：TV698.11[水利工程—水利水电工程]