摘要
考虑地下工程的复杂性和洞室围岩变形的非线性特点,提出了一种利用遗传算法优化支持向量机参数的进化支持向量机围岩位移时间序列预测方法。该方法不但克服了神经网络的过学习问题,也解决了支持向量机的参数选取问题。基于该方法,结合数据库技术建立洞室位移变形非线性时间序列计算机智能分析系统。将该系统用于水布垭电站地下厂房交通洞监测位移分析,获得满意的预测精度。
Considering the complexity of underground engineering and nonlinear displacement time series of cavern surrounding rocks,a new evolutionary support vector machine based on generic arithmetic to forecast displacement time series is proposed.Which has overcome the express learning problem of ANN and solved the parameters selection problem of SVM.Based on the method and combining database technologies,the computer intelligent analyzing system has been constructed.The system was used in the monitoring displacements analyzing of the traffic tunnel of underground workshop of Shuibuya power station,which has obtained satisfied forecast precision.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2006年第z1期808-810,共3页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(50508007)资助项目
关键词
围岩变形
时间序列预测
支持向量机
遗传算法
surrounding rock deformation time series forecast support vector machine generic arithmetic