土木工程系统辨识统计方法的现状与展望

摘要

It is important to explore statistical methods for system identification in civil engineering, because of dispersal in material properties, complexity of a structural system and randomness of environment. The system identification in civil engineering was defined here. Five statistical methods were investigated, they were Bayesian probabilistic approach, stochastic finite element method, statistical pattern recognition, statistical discrimination of features for damage detection and the other methods. The further developments of statistical methods for system identification in civil engineering were discussed based on the published investigations.%土木工程由于内在材料的离散性,结构体系的复杂性,外在环境因素的随机性,有必要引入统计方法来进行面向健康监测的损伤辨识.探讨了土木工程系统辨识的定义,分析了现有研究中与统计方法相关的Bayesian统计、随机有限元、统计模式辨识、损伤特征辨识的统计描述及其他等五类研究；根据土木工程系统辨识的特点与已有研究,展望了统计方法进一步的研究前景与方向.