针对地震预测中预测因子高度非线性、训练样本数量有限及分布不均匀的问题,提出采用RBF-BP组合人工神经网络对地震预测因子样本进行建模和预测,并用RBF神经网络初步训练预测因子样本,将训练结果输入BP神经网络,根据期望输出值加强训练。实例分析表明,该预测方法可行并能有效提高地震预测精度。 Aiming at the problem of highly nonlinear forecasting factors, limited training samples and uneven distribution of training samples, this paper proposes to use RBF-BP artificial neural network to model and predict the samples of earthquake predictor, and to use RBF neural network to train the predictor samples , The training results into the BP neural network, according to the expected output value to strengthen training. The case study shows that this method is feasible and can effectively improve the accuracy of earthquake prediction.