摘要
针对钱江四桥实时健康监测系统的监测数据,采用AR模型、ARMA模型以及灰色系统的GM(1,1)等随机模型,进行数学建模,并对各模型进行了不同步长的预测研究.结果表明,当采用合适的数学模型参数后,预测和实际监测结果十分吻合.各个模型进行比较后发现,灰色模型只需要少量原始信息,而且它的短步长预测效果优于时间序列模型;灰色模型和时间序列模型的预测误差都会随预测步长的加大而增大.
Based on the monitoring data from Qianjiang fourth bridge health monitoring system, mathematical random models including auto regressive (AR) model, auto regressive moving average (ARMA) model and grey model (GM(1,1)) were built, and the corresponding prediction results with different prediction step lengths were achieved. The prediction results by choosing appropriate model parameters agreed well with the actual monitoring results. GM(1,1) model needs less information and shows superiority to the time series model in term of short-step prediction. The prediction errors of various models increase with the increasing prediction step length.
出处
《浙江大学学报(工学版)》
EI
CAS
CSCD
北大核心
2008年第1期157-163,共7页
Journal of Zhejiang University:Engineering Science