摘要
该研究提出了一种基于主动学习的传感器优化布设新方法,并将其应用于桁架桥的变形监测。主动学习策略以逐个寻找最优传感位置的方式确定传感器的布设方案,其关键在于制定合适的准则从备选位置中选择出最优传感位置。为此,提出了一种基于最小重构误差的选择准则:首先,考虑荷载的不确定性信息,建立结构变形的高斯先验概率模型;然后,基于高斯条件分布,获取已有传感布设数据条件下的结构变形预测,下一个最优传感布设位置应使得预测协方差最小。最后,通过桁架桥仿真算例,验证所提方法的有效性,以及对复杂荷载工况、多种类型传感器的广泛适用性。
This article proposes a new method for optimizing sensor placement based on active learning and applies it to deformation monitoring of truss bridges.The active learning strategy determines the deployment plan of sensors by searching for the optimal sensing position in a one-by-one manner,and the key lies in developing appropriate criteria to select the optimal sensing position from the candidate positions.To this end,this article proposes a selection criterion based on minimum reconstruction error.Firstly,a Gaussian prior probability model of the whole structural deformation was established by considering the uncertainty information of the load.Then,based on the Gaussian conditional distribution,the full-field structural deformation was predicted under the condition of existing sensor deployment data.The next optimal sensor position was selected to minimize the covariance of the prediction.Finally,the effectiveness of the proposed method and its adaptability to complex load conditions and various types of sensors were verified through simulation examples of truss bridges.
作者
陈永亮
李亦湘
汪利
CHEN Yongliang;LI Yixiang;WANG Li(Guangxi Construction Engineering Quality Inspection Center Co.,Ltd.,Nanning 530000,China;School of Aeronautics and Astronautics,Sun Yat-sen University,Shenzhen 518107,China)
出处
《振动与冲击》
北大核心
2025年第24期182-187,277,共7页
Journal of Vibration and Shock
基金
广东省重点领域研发计划(2022B0101080001)。
关键词
传感器优化布设
主动学习
荷载不确定性
最小重构误差
optimal sensor placement
active learning
load uncertainty
minimum reconstruction error
