摘要
为了解决LMT时间序列出现的缺失和强干扰现象,根据实测资料数据量大、非线性、非平稳性等特点,首次采用ARIMA模型进行预测和填补,基于平稳性检验和贝叶斯信息准则确定模型阶数,采用最小二乘原理确定模型参数,建立双向预测模型和线性合并方法进行预测,并对比ARIMA模型和AR模型预测数据的准确度。实例表明,ARIMA模型预测结果准确,精度比AR模型高,且误差不会累积,解决了原始资料的不连续性和强干扰的问题。
To solve the deletion or skipping of LMT time series which is is large,nonlinear,non-stationary based on observed data,the ARIMA method is applied to predict the missing data and replace the jump-point.The model order number is obtained through the stationarity test and the bayesian information criteria(BIC),the model parameters is determined using the LS theory.Establishing the bi-directional prediction model and linear combination method to forecast the missing data,then compare the predicted data accuracy of ARMA and AR.The result shows that the ARMA method with higher accuracy and without error accumulation,can solve the problem that the observed data is discontinuity and strong interference.
出处
《物探化探计算技术》
CAS
CSCD
2017年第5期612-619,共8页
Computing Techniques For Geophysical and Geochemical Exploration
基金
"深地资源勘查开采"重点专项(2106YEC0600302)
国家高技术研究发计划(2014AA06A612)
