摘要
以混沌理论为基础 ,由去偏复自相关函数确定输入神经元的数目 ,通过含有隐层的误差反向传播神经网络输出预测值 ,对神经网络输出的“尖点”预测值进行混沌控制 ,通过参数控制法确定预测值 ,使预测结果更加准确。最后将该方法应用于北京市北新桥地铁车站深基坑的侧移预测 。
Based on chaos theory, this paper determines the values of input neural network by menas of the arithmetic that combining the average displacement with autocorrelation function. The prediction results are induced by the feedback network and the ‘jag points’ are reduced by the method of parameter controller. The study has shown that the results are more accurate than conventional methods. At last, the method is applied to the lateral displacement prediction of the deep excavation for Beixinqiao Subway Station of Beijing and the data are satisfactory.
出处
《工业建筑》
CSCD
北大核心
2005年第4期55-59,共5页
Industrial Construction
基金
国家自然科学基金资助项目 (批准号 :40 2 72 113 )
北京市自然科学基金资助项目 (批准号 :8992 0 0 3 )
