摘要
为了探索人工神经网络应用于地震分析预报的可能性，以福建及其周边地区地震活动为例，采用ｂ值、地震频次、地震能量释放、空间集中度４项地震活动性指标作为神经网络的输入，用具有Ｓ－型特性函数的ＢＰ网络对由每年地震的活动指标组成的标准样本进行训练，由训练结束后的权值和阈值及待预报样本的因子测值计算出网络输出值，作为地震活动性的预测．结果表明，用神经网络可以在一定精度范围内使震级预报的内检符合率达１００％，在例子中，外推预报准确率达８８％以上．
To explore the possibility of application of neural network in analysis and prediction of earthquakes, we input the following four quota of seismic activity,i.e., b value,frequency, energy release and intensity of the earthquakes in Fujian province and the areas around, and then,use BP network with S model characteristic function to exercise the standard samples coming from the annual quota of seismic activity. After the exercise, we use the weight value, the threshold value and the factorial measurement value of the sample to be predicted to calculate the network output value, basde on which we give the prediction of seismic activity. The result indicates that this neural network can,to a certain extent achieve 100% accuracy of the intrapolated earthquake prediction and 88% correctness in the extrapolated earthquake prediction.
出处
《厦门大学学报（自然科学版）》
CAS
CSCD
北大核心
1999年第4期534-540,共7页
Journal of Xiamen University：Natural Science
关键词
神经网络
BP算法
地震预报
Neural network computation, BP predication, Earthquakes
