基于MPCA-GMM的斜拉桥健康监测

Health Monitoring of Cable-stayed Bridge Based on MPCA-GMM

针对斜拉桥损伤识别问题,提出一种基于移动主成分分析(Moving Principal Component Analysis, MPCA)和高斯混合模型(Gaussian Mixture Model, GMM)相结合的方法对桥梁健康状况进行监测。以相邻测点的温度效应为基础,利用MPCA将数据进行压缩;通过GMM对压缩后的数据进行聚类分析,以参考状态平均隶属度差值95%保证率建立损伤阈值,对待测桥梁结构的健康状况进行评估。数值算例和实时监测数据分析表明该方法能有效识别结构损伤状态,并且具有良好的抗噪性。

Aiming at the problem of cable-stayed bridge damage identification,a method based on moving principal component analysis(MPCA)and Gaussian mixture model(GMM)was proposed for bridge health monitoring.Firstly,based on the temperature effect of adjacent measurement points,the data were compressed by MPCA.Then,Gaussian mixture model(GMM)was used to perform cluster analysis on the compressed data,and the damage threshold was established based on the average membership difference of the reference state with 95%guarantee rate to analyze the health status of the bridge structure under test.Numerical examples and real-time monitoring data analysis show that this method is effective in identifying structural damage state and has good anti-noise performance.