基于ARMAV模型和J-散度的结构损伤识别

Structural damage identification based on ARMAV model and J-divergence

损伤识别技术是结构健康监测系统的关键组成部分,为了进一步提高损伤识别的准确性和适用性,提出一种融合信息距离函数J-散度与向量自回归滑动平均(vector autoregressive moving average,ARMAV)模型的损伤识别方法。采用预白化过滤器对加速度时域数据进行消除激励相关性以及降噪处理;建立了ARMAV模型,并由模型的自回归参数和残差方差构建损伤判别指标;采用三层框架试验数据,并进行转播塔模型的损伤识别试验研究验证了该方法的有效性。结果表明:基于ARMAV模型和J-散度距离的损伤识别方法可操作性强,能够准确、高效地定位框架和塔架结构的损伤,且该方法受环境变化的影响较小,可为在线结构健康监测提供一种新思路。

Damage identification technology is a key component of structural health monitoring system.Here,to further improve the accuracy and applicability of damage identification,a damage identification method integrating information distance function J-divergence and vector autoregressive moving average(ARMAV)model was proposed.Firstly,a pre-whitening filter was used to eliminate excitation correlation of acceleration time-domain data,and denoise the data.Then an ARMAV model was established,and damage identification indexes were constructed based on autoregressive parameters and residual variance of the model.Finally,the effectiveness of this method was verified by using a 3-layer framework’s experimental data and conducting damage identification experiments for a relay tower model.The results showed that the proposed damage identification method has strong operability;it can accurately and efficiently locate damage to framework and tower structures;it is less affected by environmental changes;this method can provide a new idea for online structural health monitoring.