摘要
影响强震区土石坝震害的因素复杂,各影响因素也可能存在相互关联现象,两两之间为多模态且带噪声的非线性关系。利用人工神经网络具有高度容错、自学习、自组织分析复杂非线性问题的特点,结合40组国内外土石坝样本建立BP神经网络对其历史震害进行训练,使得土石坝震害与其影响因素之间的非线性关系得到较好的映射,进而建立基于BP神经网络的土石坝震害等级预测模型。实例验证表明,该BP网络能够对土石坝震害进行有效预测。
Factors that affect the earthquake disaster of earth-rockfill dam are very complex.These factors may be interrelated each other with multi-modal and non-linear noise relationship.Artificial neural network has characteristics of high fault-tolerant,self-learning,self-organization analysis and can be analyzed complex nonlinear problems.BP neural network model is established to train the historical earthquake disasters samples with 40 group data sets of earth-rockfill dam in the world.The non-linear relationsh...
出处
《水电能源科学》
北大核心
2010年第5期69-73,共5页
Water Resources and Power
关键词
土石坝
地震
人工神经网络
震害预测
精度
earth-rockfill dam
earthquake
artificial neural networks
earthquake damage forecast
precision
