【摘要】 研究地震测试,为了精确预报地震,减少伤亡和捷足损失,针对地震预测中的预测因子数据高度非线性问题,以及训练样本数量有限和分布不均匀的问题,给测试带来困难。采用组合人工神经网络对地震预测因子样本进行建模和预测。用自组织映射(SOM)神经网络对地震预测因子样本进行分类,使原本较为分散的样本点各自集中到内在规律较为相似的样本类中。再在各样本类中分别构建径向基函数(RBF)神经网络进行训练和预测。通过对地震实例的预测仿真及其相关分析表明,预测方法具有对地震预测的可行性。并且优于传统的BP神经网络的地震预测方法,方法可以有效提高预测精度。更多还原

【Abstract】 Considering the limitation and unevenly distribution of the number of samples and the highly non-linear problem of earthquake prediction,a combined artificial neural network method is adopted in this paper.There are two steps in this method: clustering the samples by using SOM neural network at first,and then training and predicting at the clustered areas by using RBF neural network,respectively.Because the samples have the similar inherent regularity after SOM neural network classification,thus,the prediction accuracy of RBF neural network is improved.The simulation results show that the proposed method is an effective tool for the prediction of earthquake,and it can effectively enhance the prediction accuracy by compared with using BP neural network.更多还原