Applied Research Of BP Neural Network In Earthquake Prediction

Abstract:

Nowadays, earthquake prediction is still a worldwide scientific problem, especially the prediction for short-term and imminent earthquake has no substantial breakthroughs. BP neural network technology has a strong non-linear mapping function which could better reflect the strong non-linear relationship between earthquake precursors and the time and the magnitude of a potential earthquake. In this paper, we selected the region of Beijing as the research area and 3 months as the prediction period. Based on BP neural network and integrated with the conventional linear regression method, a regional short-term integrated model was established, which gives the quantitative prediction for the earthquake magnitude. The results show that the earthquake magnitude prediction RMSE (root mean square error) of the integrated model reaches ± 0.28 Ms. Compared with conventional methods, the integrated model improves significantly. The new model has a good prospect to use BP neural network technology for earthquake prediction.

• Periodical: Applied Mechanics and Materials (Volumes 204-208)
• Pages: 2449-2454
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.204-208.2449
• Online since: October 2012
• Authors: Wu Sheng Hu, Hong Lin Nie, Hao Wang
• Keywords: BP Neural Network (BPNN), Earthquake Parameter, Earthquake Prediction, Linear Regression, Time-Varying Parameter

References

[1] Geller J Robert, David D Jackson, Yan Y Kagan, et al. Earthquakes Cannot Be Predicted [J]. Science, 1997,275(5306): 1616-1617

DOI: 10.1126/science.275.5306.1616

[2] Knopoff L. Earthquake prediction is difficult but not impossible. Nature debate on earthquake prediction,11 March,(1999)

DOI: 10.1038/nature28113

[3] Aki K. A new view of earthquake and volcano precursors [J]. Earth Planets and Space, 2004,56(8): 689-713

DOI: 10.1186/bf03353079

[4] Han Zhiqiang, Wang Biquan. Short-term Earthquake Prediction Method Based On Neural Network [J]. Earthquake Science,1997,19(4): 367-375

[5] Han Zhiqiang, Wang Biquan. Short-term Earthquake Prediction Method Based On Neural Network [J]. ACTA SEISMOLOGICA SINICA, 1997, 19(4): 367-375 (in Chinese)

[6] Cai Huangdong, Gan Junren, Yao Linshen. Application of Artificial Neural Network to Synthetic Prediction of Earthquakes [J]. Earthquake Science,1993,15(2): 257-260

[7] Wang Wei, Jiang Chunxi, Zhang Jun, et al. The Application of BP Neural Network To Comprehensive Earthquake Prediction [J]. Earthquake,1999,19(2): 118-126 (in Chinese)

[8] Hu Wusheng. Neural Network Theory And Its Application [M]. Beijing: Sino Maps Press,2006. 63-83 (in Chinese)

[9] Lu Yuanzhong, Li Shengle, Deng Zhihui, et al. Seismic Analysis Forecast System Based On GIS [M]. Chengdu: Chengdu Map Press. 2002. 20-30 (in Chinese)

[10] Ren Zhenqi. The Historical Earthquakes In Beijing Region[J]. Recent Developments in World Seismology, 1996, 12(9): 34-35 (in Chinese)

[11] Pan Bo, Xu Jiandong, Haruko Sekigguchi, et al. Simulation Of The Near-Fault Strong Ground Motion In Beijing Region [J]. Selismology and Geology, 2006, 28 (4): 623-634 (in Chinese)

[12] Wang Suyun, Yu Yanxiang. Research on Empirical Relationship of Earthquake Magnitude Scales and Its Influence on Seismicity Parameters[J]. Technology for Earthquake Disaster Prevention 2009, 4(2): 141-149 (in Chinese)

[13] Wang Wei, Dai Weile, et al. The Forecasting Earthquake Method of Neural Networks in the Short term Using Anomalous Data of M f, C and D Values[J]. Earthquake Research In China, 1997, 13(4) (in Chinese)