摘要
运用人工神经网络原理 ,对BP型神经网络作了多方面的改进。采用改进后的BP算法 ,建立了建筑物震陷预测模型。研究结果表明 ,改进的BP网络性能良好 ,所建立的模型预测精度高 ,能满足工程要求 。
Based on the principle of artificial neural networks,the error back-propagation(BP)algorithm is improved in some aspects.And using the improved algorithm,a model is established for the prediction of building settlements due to earthquake liquefaction.The result show that the developed BP model is advanced,can satisfy the requirement of engineering in practice and is a new effective method for prediction.
出处
《地震研究》
CSCD
北大核心
2001年第3期262-266,共5页
Journal of Seismological Research
基金
广东省自然科学基金 (990 14 8)
广东工业大学基金 (982 0 33和 2 0 2 16 )资助
关键词
建筑物震陷
预测
动态自适应BP网络
地基失效
人工神经网络
Building settlements due to earthquake liquefaction
prediction
dynamic selfadjustment error back-propagation network
failure of foundation
