Real-time stress monitoring data can reflect the security state of engineering structure. For the potential characteristics of periodic oscillation along with the change of temperature in the stress time series which are acquired from the first steel anchor box of Sutong Bridge,singular spectrum analysis is used and improved to extract the trend components and cycle components from original sequence. Then the good analysis and prediction ability of ARIMA model is applied to forecast the principle trend and cycle components. Finally,the predictions will be added and the sum should be corrected. Compared with the traditional ARIMA model and the PSO-SVM(support vector machine optimized by particle swarm optimization)model,the results prove that the predictive sequences produced by the proposed model are most close to the observed ones. Therefore,the method proposed above has certain application value in the health monitoring of engineering structural stress.%工程结构的实时应力监测数据可以反映其自身的安全状态。针对苏通大桥北索塔锚固区下部首节钢锚箱的应力时间序列具有潜在的随温度呈周期振荡变化的特征，引入奇异谱分析方法，并加以改进，从原序列中提取出趋势成分和周期成分。再利用ARIMA模型对振荡序列有着良好的分析预测能力，对趋势及周期主分量进行预测，最后对分项预测结果加和校正。结果表明，与纵向的传统ARIMA模型及横向的基于粒子群优化算法的支持向量机（PSO-SVM）模型相比，上述所提方法的预测结果最为接近实测序列，该模型在工程结构应力健康监测中具有一定的应用价值。
展开▼
