【摘要】 研究了将自适应局部线性化方法应用于较强余震间隔时间的预测。首先对余震间隔时间数据进行预处理;再利用自适应局部线性化法找出合适的嵌入维数,并计算出待估计参数。最后用待估计参数预测余震间隔时间。对汶川地震发生后4.0级(含4.0级)以上的余震间隔时间数据进行了预测。实验结果表明,自适应局部线性化方法预测的余震间隔时间的平均均方根误差和平均绝对偏离明显低于标准局部线性化方法;自适应局部线性化方法预测的余震间隔时间绝对误差比标准局部线性化方法和最小二乘拟合方法的预测结果分别减少了7 d和20 d.更多还原

【Abstract】 The approach to predicting time interval between aftershocks using adaptive local linear(ALL) is studied.Firstly,the data of time interval between aftershocks should be preprocessed.Then suitable embedding dimension is found using ALL,and the parameters to be estimated are calculated.Finally,time intervals between aftershocks are predicted by the parameters.The datasets of time intervals between aftershocks measuring above 4.0-magnitude(including 4.0 magnitude) after the Wenchuan earthquake are used to predict.Experimental results show that MRMSE(mean of root mean square errors) and MMAE(mean of mean absolute errors) of predictive time interval using ALL are much less than that of using original local linear(LL).The AE(absolute error) of predictive time interval using ALL decreases by 7 days and 20 days respectively than that of using LL and least square fitting.更多还原