因子相关性对两种大坝监测模型的影响分析

Analysis of effects of factor correlativity on two kinds of dam monitoring models

通过实例分别对BP神经网络模型和统计回归模型进行了建模因子的主成分分析,通过对相应原始模型的比较,研究了因子相关性对两种模型的影响,结果证明因子相关性对BP神经网络模型基本无影响,对统计回归模型影响较大.因此,对效应量做预报时,采用BP模型或主成分回归模型比较合适,因为它们不受因子相关性影响;如果要分析效应量和自变量之间的关系,考虑BP模型难以用明确的方程式表示,则采用主成分回归模型较为合适.

The principal component analysis of factors of BP neural network model and statistical regression model has been carried out by an example; and the effects of factor correlativity on the two kinds of dam monitoring models are studied. The result shows that the factor correlativity doesn't affect BP neural network model; but affects the statistical regression model greatly. Finally, the possible use conditions of two kinds of model are given.