自回归模型在井水埋深预测中的应用及改进

用时间序列分析方法建立井水埋深的预测模型时,首先采用差分的方法把季节性时间序列变成平稳时间序列,在此基础上,再用动态数据系统方法的传统F检验定阶法进行分析。由于样本的随机性可能过早地退出对模型的循环检验,从而不能找到合适的预测模型。笔者在用自回归模型建立井水预测模型的基础上,采用了一种改进的建模方法,提高了预测精度,并用实例进行了验证。

Application and Improvement of Auto-regression Model in Prediction of Well Depth

When time sequence analysis method is used to establish prediction model of well depth, the seasonal time sequence values are first transformed into steady time sequence values by the differences in seasonal time sequence values followed by analysis of traditional F-test rank determination based on DDS method. Because of the random characteristics of the sample, appropriate prediction model could not be found due to too early withdrawing of the cycling test for the model. Based on the establishment of auto-regression prediction model of well depth, an improved model is adopted to raise the precision of prediction. The method is verified with realistic example.