摘要
在地震地质背景研究的基础上,将三峡库首区划分为31个预测单元,再根据不同的诱震组合条件组合成35种方案,选用8个影响因子,运用神经网络理论中改进的BP算法对三峡水库诱发地震强度进行了预测。预测结果对比表明,应用神经网络模型预测的三峡库首区水库诱发地震的结论是可靠的,与其它的数学预测模型相比,运用神经网络模型预测水库诱发地震,具有能充分利用所掌握的资料,能充分表达变量间复杂的非线性关系以及预测精度较高等优点,值得推广应用。
On the base of research on earthquake geological background,the head area of TGP reservoir is divided into 31 prediction units,and in the light of 35 different composite earthquake inducing conditions,the induced earthquake intensity in TGP reservoir area is predicted using 8 influence factors by improved BP algorithm of neural network theory.The prediction result demonstrates that the conclusion of induced earthquake in TGP reservoir head area by neural network model is reliable and more accurate.Compared with other mathematical prediction models,using neural network model to predict induced earthquake by reservoir can more fully make use of data and express complicated non-linear relation of variables,possesses advantage of high prediction accuracy,deserving popularization.
出处
《人民长江》
北大核心
2007年第2期8-10,共3页
Yangtze River
基金
国家重点基础研究发展计划(973计划)资助项目(2004CB720200)
关键词
水库诱发地震
神经网络
影响因子
震级
预测
三峡水库
induced erathquake by reservoir
neural network
influence factor
earthquake magnitude
predicion
TGP reservior
