论文部分内容阅读
将系统分析方法中的传统GM(1,1)模型与时间序列分析方法相结合建立地下水位的预报模型.为更好地反映地下水位的统计规律随时间而缓慢变化的现象,采用衰减记忆最小二乘法率定GM(1,1)模型参数.用方差分析法进行地下水位时间序列的周期分析时,对初相进行优选;在选定的置信度水平下,当有多个周期都通过检验,此时该如何优先选择哪个周期,文中提出按拟选周期的F分布检验统计量与检验区间界限值的比值最大原则来选择周期成分.最后,采用本文所述方法进行福建省龙岩市的年最高地下水位的预报.
The forecasting model of groundwater level is established by combining the traditional GM (1,1) model and time series analysis method in system analysis method.In order to better reflect the phenomenon that the statistical law of groundwater level slowly changes with time, the minimum attenuation memory (1,1) model parameters by using the second-order multiplication method.When analyzing the periodicity of groundwater level time series by ANOVA, the initial phase is optimized.When there are multiple cycles at the selected level of confidence, In this paper, we propose to choose the period according to the maximum principle of the ratio of the F distribution test statistic to the test interval limit of the selected period.Finally, using the method described in this paper, Forecast of the highest groundwater level.