面向桥梁结构状态预测的ARMA-GM组合时序模型研究

Study on ARMA-GM Combined Time-series Model for Structural State Prediction of Bridges

桥梁健康监测数据中蕴含引起桥梁结构状态变化的信息,通过分析其特征变化而实现结构状态预测的方法已得到工程界和学术界的重视。为了克服单一时间序列模型ARMA和灰色关联模型GM(1,1)在桥梁结构状态预测中的不足,提出一种ARMA-GM组合时序预测模型,以描述监测数据序列前后之间的数学关系,并对未来某一时间段内的监测值进行预测。实验结果表明:组合模型在预测步长增大时预测的平稳性好,而且比单一模型的预测精度更高,能够为桥梁结构安全状态评估提供宝贵的预测数据。

The bridge health monitored data contained the information indicating the structural state changes of bridges. The method for structural state prediction by analyzing the data feature variations has drawn the attention in engineering and academic field. To overcome the weakness of single time-series model of ARMA and Grey-relation model of GM （ 1,1 ） in the structural state prediction, a combined time-series model of ARMA-GM was proposed to describe the mathematical relationship between the former and later monitored data series, and to achieve the prediction of the values monitored in a future period. The experiment results show that the proposed combined model demonstrated better stability and higher prediction accuracy than the single model when the prediction step is lengthened. Consequently the proposed combined model can provide the valuable predicting data for structural safety assessment of bridges.