| 神经网络方法在地震预报研究中的初步应用 |
| |
| 引用本文： | 冯德益,汪德馨.神经网络方法在地震预报研究中的初步应用[J].地震,1994(4):23-29. |
| |
| 作者姓名： | 冯德益 汪德馨 |
| |
| 作者单位： | 天津市地震局，天津市地震局，天津大学，天津市地震局，天津市地震局 中国天津 300201，中国天津 300201，中国天津 300072，中国天津 300201，中国天津 300201 |
| |
| 摘 要： | 本文把神经网络方法引进地震预报研究当中。使用地震频次，最大震级，平均震级，等价地震次数等多项地震活动性指标作为神经网络的输入，未来时段内的最大地震震级作为其输出，可以对某一固定地区的最大地震震级作出中近期预报。选用的神经网络模型为含两个中间层的前向模型，并采用ＢＰ算法。所得结果表明，用神经网络方法可以在一定精度范围内使震级预报的内检符合率达到１００％，在本文的例子中，外推预报准确率达到６０％以上。
|
| 关 键 词： | 神经网络 前向模型 地震预报 |
PRELIMINARY APPLICATION OF NEURAL NETWORK METHOD IN EARTHQUAKE PREDICTION RESEARCH |
| |
| Feng Deyi Jiang Chun.PRELIMINARY APPLICATION OF NEURAL NETWORK METHOD IN EARTHQUAKE PREDICTION RESEARCH[J].Earthquake,1994(4):23-29. |
| |
| Authors: | Feng Deyi Jiang Chun |
| |
| Institution: | Feng Deyi Jiang Chun
(Seismological Bureau of Tianjin)
Wang Dexin
(Tianjin University)
Tian Shan Zheng Ximing
(Seismological Bureau of Tianjin) |
| |
| Abstract: | This study has introduced the method of neural networks into the earthquake prediction reseach.By using several seismicity indecies, such as earthquake frequency, maximum magnitude, average magnitude, and the frequency of equivalent earthquakes as the input and the maximum magnitude in forthcoming time interval as the output of neural network, the magnitude of largest earthquake in the given region in near period ( for example, 1 year ) may be predicted. The forward-type model of neural network containing two medium layers is selected and the BP algorithm is used.The result shows that by use of the method of neural networks the consistency degree in the magnitude prediction in certain accuracy may reach 100% in the internal examinations and above 60% in the extrapolated prediction tests. |
| |
| Keywords: | neural network forward-type model BP algorithm earthquake prediction |
| 本文献已被 CNKI 维普 等数据库收录！ |
