高土石坝短序列监测资料建模研究

Modeling Study of the Short Sequence Monitoring Data of High Earth-rockfill Dams

通过对逐步回归模型、灰色GM(1、N)模型,BP神经网络模型3种典型建模方法原理的研究比较,再结合对大量具体高土石坝短序列资料建模效果的比较,提出了高土石坝短序列资料的分类,并引进了关联度作为分类标准,给出了不同类型适宜采用的模型,最后通过对冶勒沥青混凝土心墙坝短序列资料的具体计算,证明了分类合理,值得推广。

Based on comparing and researching into three typical modeling methods stepwise regression,grey GM（1 ,M） and B-P neural networks, combined with comparing the modeling effect of a large number of high earth-rockfill dams＇ short sequential monito ring data, the authors advanced the classification of the high earth-rockfill dam＇s short sequences, introduced the relational grade to the classification standard and give the suitable modeling method for different type. Finally, through the factual modeling analysis of the YEI.E asphalt concrete core wall earth-rockfill dam , it was proved to be reasonable and worthy of popularization.