基于混沌神经网络理论的城市地面沉降量预测模型  

CHAOS NEURAL NETWORK THEORY BASED MODEL FOR QUANTITATIVE PREDICTION OF URBAN GROUND SUBSIDENCE

作　　者：李峰[1] 宋建军[1] 董来启[1] 张先哲[1] 武艳丽[1] 张玲 

机构地区：[1]河南省煤田地质局资源环境调查中心,郑州450002 [2]河南省煤炭地质勘察研究院,郑州450052

出　　处：《工程地质学报》2008年第5期715-720,共6页Journal of Engineering Geology

摘　　要：通过分析城市地面沉降量时间序列的非线性动力学系统,认为该时间序列具有混沌特性。在此基础上,通过相空间重构的方法建立了用于城市地面沉降量预测的混沌神经网络模型;并利用此模型对高桥地面沉降量进行了预测,并和实际监测沉降量进行了比较,最大绝对预测误差为1.7,预测的平均误差为0.0833,研究结果表明,应用混沌神经网络模型进行城市沉降预测是可行、精确的。urban ground subsidence is of nonlinear dynamic character and its quantity in time series is analyzed in this paper. Then it is assumed that there is chaos in the urban ground subsidence in time series. Based on this assumption and using the chaos neural network theory, a prediction model of urban ground subsidence quantity was built with phase space reconstruction. The ground subsidence quantity in Gao - qiao analyzed and predicted with this model. The observed data are compared with the predicted data. The largest absolute prediction error is 1.7 and the average forecast error is 0. 0833. The results indicate that the chaos neural network theory is reasonable and accurate to predict the urban ground subsidence.

关 键 词：城市地面沉降 混沌 时间序列 混沌神经网络 高桥 

分 类 号：P642.26[天文地球—工程地质学]