摘要
为提高时间序列分析法在地下水位预报中的准确性，本文对时间序列分析法进行了改进。调整随机项的自回归系数，使其代数和为1，可减小“蝴蝶效应”对预报期后段结果产生的影响，从而提高预报结果的准确性。通过对镇赉县10口井2000-2009年5日地下水位数据进行分析与检验，结果表明：用传统法检验结果的相对误差范围是8.8％～21．9％，而应用改进法检验结果的相对误差范围是2．1％～10.3％，其中6口井的相对误差提高了10％以上。模型经改进后，拟合精度得到了显著提高。改进法可以更有效地对地下水位动态进行预报。
To enhance the accuracy of groundwater level, this study makes some improvements on time se- ries analysis. To adjust the autoregressive coefficient of random item helps to avoid ＂The Butterfly Effect＂, which makes the forecast results independent. By analyzing and testing the groundwater level data observed every 5 days from the year of 2000 to 2009 in Zhenlai county, the relative errors of improved method vary within 2.1%-10.3%, which varied from 8.8% to 21.9% in traditional method. Furthermore, the rela- tive errors are increased more than 10% in 6 wells. The fitting precisions have been improved significantly by improved method. In conclusion, improved method is an effective method in forecasting groundwater level.
出处
《水利学报》
EI
CSCD
北大核心
2013年第11期1372-1379,共8页
Journal of Hydraulic Engineering
基金
教育部博士点基金项目(20120061110058)
吉林省科技厅科技发展计划项目(20130206011SF)
国家自然科学基金项目(41072171)
