基于PCA-BPNN的膨胀土自由膨胀率预测

Free swelling ratio prediction of expansive soil based on PCA-BPNN

为了准确地对膨胀土进行分类,将主成分分析方法(PCA)和BP神经网络方法(BPNN)相结合,进行膨胀土自由膨胀率预测研究。首先,对影响自由膨胀率的几个物性指标数据进行主成分分析,得到主成分数据序列;然后,根据主成分数据序列建立输入样本并应用BP神经网络进行训练,得到PCA-BPNN预测结果,从而预测自由膨胀率。采用南阳膨胀土数据进行实例分析,结果表明,该模型比单一BP神经网络预测方法的效果更好。

In order to classify swelling soil accurately,we conduct study on free swelling ratio prediction of expansive soil by combining Principal Component Analysis（PCA） with BP Neural Network method（BPNN）.First,we obtain PCA data sequence with several physical index data that affects free swelling ratio of expansive soil;then,according to PCA data sequence,we establish input data.We apply the data to the BPNN training and get prediction results of PCA-BPNN,thus predict free swelling ratio.We analyze an example by adopting the swelling soil data from Nanyang city,the results show that the model is much better than a single BPNN forecast method.