摘要
一般情况下,基坑工程位移的发展趋势可以分为3种类型,传统GM(1,1)模型由于模拟曲线为指数曲线,因此只适合于第1类趋势的位移时间序列建模。在无偏GM(1,1)模型的基础上建立修正曲线型无偏灰色预测模型和生长曲线型无偏灰色预测模型,适合于第2、3类趋势的位移时间序列建模。结合神经网络的非线性描述能力以及无偏灰色预测模型的趋势预测能力建立神经网络误差修正灰色模型。基坑位移预测实例应用结果显示,神经网络误差修正灰色模型能很好地描述基坑位移的非线性发展。
The displacement development trend of foundation pit can be divided into three types.Because the simulation curve is an exponential one,the existing GM（1,1） model is only suitable to establish the model for the data sequences of first development trend type.Based on the unbiased GM（1,1） model,the unbiased grey forecasting model of modified exponential curve and unbiased grey forecasting model of growth curve are proposed to establish models for displacement-time sequence of the second and third displacement trends.Combining non-linear description function of neural network and development trend prediction function of unbiased grey model,the error-corrected grey model based on Neural Network is proposed.The practical application of foundation pit displacement forecast using the established model shows that the model can describe the non-linear development trend of displacement.
出处
《人民长江》
北大核心
2010年第17期25-29,共5页
Yangtze River
关键词
基坑
无偏灰色预测模型
神经网络
误差修正
位移预测
foundation pit
unbiased grey forecasting model
neural network
error correction
displacement forecast
