针对结构健康监测中如何利用在线监测数据进行损伤诊断的问题,基于时间序列分析提出了一种新的损伤预警方法。首先对所有监测数据样本建立ARMA模型,以模型中AR部分参数的土成分矩阵构建Mahalanobis距离判别函数,进而提出了一种新的结构损伤敏感特征指标DSF。然后,采用t检验考察该指标的均值在损伤前后是否存在显著性变化,从而可以实现结构的损伤预警。最后,对Benchmark结构在环境激励下的试验数据进行了损伤预警研究。结果表明:该文基l丁时序方法提出的DSF对结构的微小损伤具有敏感性,具备在线实时损伤预警的应用价值。
A time-series-based damage alarming method is presented to solve the problem of on-line damage diagnosis in structural health monitoring (SHM). Auto-regressive moving-average (ARMA) models were firstly built according to the monitoring data, and then a principal-component matrix was derived from the AR coefficients of these models to establish the Mahalanobis distance criterion functions. After that, a new damage sensitive feature index DSF was proposed. A hypothesis test involving t-test method was carried out to determine if damage alarming is needed by checking whether or not the mean value of DSF had significant changed after damage. At last, the ambient vibration test of the IASC-ASCE Benchmark structure was taken as a case study. Results show that, the time-series-based index DSF is sensitive to structural minor damage, and the proposed method can be applied in the on-line damage alarming in SHM.