偏最小二乘回归神经网络模型在大坝观测资料分析中的应用

APPLICATION OF NEURAL NETWORK MODEL WITH PARTIAL LEAST-SQUARES REGRESSION INTO INTERPRETATION ON OBSERVATION DATA OF DAM

偏最小二乘回归方法能较好地解决自变量之间的严重相关性问题,神经网络模型可以克服模型必须是基本观测数据的线性和非线性组合的局限。研究将两者有机地联系起来,建立偏最小二乘回归神经网络模型,以克服上述不足。实例分析表明,本模型预报精度较高。

Partial least-squares regression method can resolve the problem of serious multicollinearity among variables, and the neural network can overcome the shortcoming that the conventional models must be the combination of linearity and non-linearity of input data. The two methods, neural network model with partial least-squares regression of observation data of dam are combined. The result of an example shows that the prediction is of high precision with the proposed method.