用一种新的方法对避免共线性的影响进行探讨 ,建立了两个人工神经网络模型 ,得到了较好的拟合效果 ,同时进行了参数分析。神经网络模型为建立存在共线性现象的多元线性回归模型开辟了新的途径。 A new method was used to explore the effect of avoiding collinearity. Two artificial neural network models were established, the better fitting effect was obtained and the parameters were analyzed at the same time. The neural network model opens up a new way for the establishment of multivariate linear regression models with co-linear phenomena.