Research Of Artificial Intelligence Damage Recognition Based On Localized Time-dependent Reliability Of Bridges

The basic connotation of Bridge Health Monitoring is to provide early warning signals for bridges in special climates,special traffic conditions or when the operation of bridges is very serious,and to provide basis and guidance for maintenance and management decision-making of bridges.Bridge Health Monitoring is not only a new technology of traditional bridge detection and structure evaluation,but also has been endowed with three aspects of significance: structure monitoring and evaluation,design verification and research and development.Nowadays,many new bridges and long-span bridges at home and abroad have been equipped with bridge health monitoring system.The powerful data acquisition ability of the monitoring system records a large amount of monitoring data.However,due to the lack of in-depth research on monitoring data analysis and mining,the massive health monitoring data can't be properly processed and utilized,which limits the monitoring system to play its role of scientific guidance.In view of the problems of inadequate research on long-term health monitoring data of bridges and inadequate processing and utilization of massive monitoring data,this paper focuses on the application of long-term health monitoring data and carries out the following work:(1)The long-term health monitoring temperature data is applied to the statistical study of temperature effect of concrete box girder.The temperature field distribution characteristics,temperature gradient and representative value of temperature difference of concrete box girder are determined by statistical analysis method,and compared with the recommended value of the code.(2)The strain data of long-term health monitoring are applied to the study of localized time-dependent reliability analysis of bridges,and the calculation method of localized time-dependent reliability of bridges based on long-term health monitoring data is improved and revised.Cluster analysis is used to determine the statistical time division method of load effect of Zhaoqing West River Bridge;hypothesis test is used to determine the statistical distribution type of stress data at the measured points of Zhaoqing West River Bridge;and the reliability and safety of a measured point of Zhaoqing West River Bridge are calculated by using the proposed localized time-dependent reliability analysis method.(3)Taking the simply supported plane beam model as the research object,by means of the large general finite element software ANSYS,the long-term operation state of the bridge is simulated by applying the measured temperature data on the model,the sample bank of localized time-dependent reliability index is established,the variation rule of reliability index with damage degree is determined,and a damage identification method for a simply supported plane beam is proposed using local reliability index as machine learning input.