关键词: 浸润线预测, 灰色神经网络, 欧拉公式, 组合预测模型

Abstract: During the process of calculation, the problem of GM(1,1) algorithm that used to solve the development factor and the gray affections can be influenced by the background value. The identifying parameters method of avoiding background value was proposed for calculation. The problem of low prediction accuracy caused by the background value of the selected steps or improper selection was solved. The problem of initial condition with fixed value can affect the prediction accuracy when using the GM(1,1) model predicts. The method of constructing the initial value with the variable weight was proposed. The impacts of the fixed initial value to the prediction accuracy was avoided. Aiming at the problem of single sample type of traditional grey neural network, a new combination forecasting model was proposed. To tackle the limitation of the traditional model to predict saturation line only rely on the data of history single sample type, and the prediction model of saturation line based on improving grey neural network is established. Through engineering verification, it is shown that the model effectively predict the change of the saturation line height in the short term.

Key words: saturation line prediction, grey neural network, Euler's formula, combination prediction model