桩筏(箱)基础沉降多步预测控制的IPSO-Elman算法  

Multi-step prediction control for IPSO-Elman combining algorithm with settlement of pile-raft(box) foundation

作　　者：郭健[1] 王元汉[1] 苗雨[1] 向平[2] 

机构地区：[1]华中科技大学土木工程与力学学院 [2]华中科技大学控制工程与科学系,湖北武汉430074

出　　处：《华中科技大学学报（自然科学版）》2008年第6期96-99,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：中国科学院武汉岩土力学研究所重点实验室开放课题资助项目(Z110507)

摘　　要：将改进的粒子群优化(IPSO)算法与Elman神经网络进行了有机结合,形成了IPSO-Elman混合算法.建立桩筏(箱)基础沉降变形期望输出与超前预测输出之间的非线性隐式方程,避开了复杂的岩土工程本构关系和力学参数计算问题.提出的多步预测控制方法,具有很好的全局识别特点和较高的推广预测能力.工程实例分析表明,IPSO-Elamn算法在桩筏(箱)基础沉降的非线性系统动态辨识和在线预测应用方面,具有良好的预测精度,满足工程实际需要.Based on improved particle swarm optimization （IPSO）, and the study of the nonlinear-dynamic identification of Elman neural network, a new algorithm was presented, in which IPSO algorithm was combined with the Elman network. The nonlinear implicit equations were constructed between target and prediction of the settlement of pile-raft （box） foundation. It was proved that this approach can be avoided the complicated structure relation and mechanical parameters in geotechnical engineering. Multi-step predictive method reveals a good characteristic of global optimization and a better performance in prediction. The results of the example show that the method is valuable. It is suitable for dynamic identification of nonlinear system and on-linear prediction. The precision of prediction can meet engineering need.

关 键 词：桩筏基础沉降 改进粒子群优化算法 ELMAN神经网络 动态辨识 多步预测 

分 类 号：TU375[建筑科学—结构工程]