摘要
针对地下工程变形监测中位移单调增长的特性 ,根据位移分解原理 ,采用灰色系统提取位移序列的趋势 ,用基于免疫进化规划的进化神经网络逼近偏差序列 ,提出了变形预测的集成智能建模方法 ,很好地解决了地下工程变形预测的难题。两个工程实例的实测位移序列建模预测研究结果证明了提出的集成智能预测方法的有效性及可行性 ,其结果显示出集成智能方法不但能得到良好的位移逼近效果 ,而且具有良好的预测泛化能力。
Considering on the monotonously increasing character of the displacement series in underground engineering, a new integrated intelligent method for displacement prediction is proposed. This method extracts the trend using Grey System model and analyzes the deviation using the ENN whose architecture and algorithm parameters evolved simultaneously from the principles of displacement decomposition. And the efficiency and the feasibility of this method is verified by two engineering examples.The results show that performances of displacement approach and displacement extension prediction of the method are all good.
出处
《土木工程学报》
EI
CSCD
北大核心
2004年第4期87-92,共6页
China Civil Engineering Journal
关键词
地下工程
位移序列
变形预测
集成智能方法
underground engineering
displacement series
displacement prediction
integrated intelligent method
