论文部分内容阅读
用混合回归分析方法既可以研究因变元和自变元之间的统计依赖关系,亦可以研究因变元自身的前后统计依赖关系。对大灰厂、牛口峪在唐山地震前后的短水准资料、降雨和温度资料分别进行了混合回归分析计算,用近似AIC准则判定最佳疏系数混合回归模型,其残差量缩小约6—15倍左右。疏系数混合回归模型各延迟变元排列顺序的变化表明,地震前后模型的统计特征发生了明显变化,由此可划分地震前的正常期、地震时的异常期和地震后的正常期。所以,此方法可提取形变资料中的地震异常信息,是分析形变观测资料的一种新方法。
Using the method of mixed regression analysis, we can study the statistical dependence between the dependent and the dependent variables, and also study the statistical dependence of the dependent variables themselves. The short-term data, rainfall and temperature data of the large ash factory and Niutu valley before and after the Tangshan earthquake were respectively analyzed by mixed regression analysis. The approximate regression AIC criterion was used to determine the optimal sparse coefficient mixed regression model. The residuals were reduced by about 6- About 15 times. The variation order of the delayed variables in the mixed-sparse coefficient regression model shows that the statistical characteristics of the model before and after the earthquake have changed obviously, which can be divided into the normal period before the earthquake, the anomalies during the earthquake and the normal period after the earthquake. Therefore, this method can extract the seismic anomaly information in the deformation data, which is a new method to analyze the deformation observation data.