摘要
本文通过对福建及其周边地区地震活动人工神经网络模型的构建 ,研究了人工神经网络方法在基于该区域地震活动性指标的地震分析预报中的应用。选用含一个中间层的前向神经网络模型 ,并采用与之相适应的BP算法 ,以该地区1 971~ 1 997年的地震活动性资料为基础 ,用神经网络进行实际计算、分析和检验。结果表明 :神经网络模型对福建及其周边地区地震震级的预测检验效果较好 ,可以在一定精度范围内使震级预测的内符率达 1 0 0 % ,外推预报准确率达 90 %。
This paper tries to build the neural networks method of seismic activity in Fujian and its adjacent area .The application of the neural networks method in earthquake prediction based on the indices of seismic activity in this region have been studied.The forward type model of the neural networks containing one medium layer was adopted,and the BP algorithm was taken correspondingly .The data which were obtained from the earthquakes activity indices in Fujian(1971~1997) are used to train,evaluate and analize the neural networks.The results show the neural networks has good quality for pattern recognition,and is suitable for earthquake prediction to Fujian and its adjacent area,by use of the methods the consistency degree in magnitude prediction in certain accuracy may reach 100% in the interal examinations and above 90% in the extrapolated prediction tests.
出处
《台湾海峡》
CAS
CSCD
2000年第1期107-112,共6页
Journal of Oceanography In Taiwan Strait
关键词
人工神经网络
BP算法
预报模型
地震活动
Neural networks, BP algorithm,indices of seismic activity earthquake,prediction method