Research On Damage Neural Network Method Of Arch Bridge Suspender Based On Deflection Monitoring

Derrick is an important component of the arch bridge,in the course of normal service,the arch bridge is affected by the long-term dynamic load and environmental erosion that lead to the anchorage failure and fracture of the derrick frequently,so it is great significant to carry out real-time and accurate health monitoring.The study proposes a neural network identification method for derrick damage based on deflection monitoring,the main research contents of this paper include:(1)The BP(Back Propagation)neural network and the RBF(Radial Basis Function)neural network are studied and analyzed,and the two neural networks are improved and the parameters are adjusted respectively.RBF neural network is chosen as the other means of derrick damage through the comparison of fitting time and the mean-square error.(2)In this paper,we study the increase of deflection change law of anchorage point of tie beam under uniform load when derrick is damaged,and establish a neural network with the increase of deflection increment of each anchorage point of the main beam as the input,and optimization of extension constants,recognition ranges,sample sampling methods,number and location of sensors.The results show that under the action of uniform load,the neural network has a large damage degree and a large identification error of side span derrick,but the network can still perform accurate identification of the damaged derrick in the most unfavorable identification situation.(3)Studying the increase of deflection change law of tie beam’s anchorage point under the combined action of uniform load and overall lifting temperature when derrick is damaged,combined with the conclusions in uniform load,the optimal temperature index data is obtained,and the damage identification network of the derrick under the combined action of the temperature load and the overall lifting temperature is optimized.The results show that the neural network can still accurately identify the derrick damage under the combined action of uniform load and overall lifting temperature,and the identification accuracy is improved.(4)Studying the increase of deflection change law of tie beam’s anchorage point under the the action of moving load and uniform load when derrick is damaged,Analysis of the influence of the number of sensors and the loading position of the monitoring vehicle on the recognition accuracy of the network,and optimization of the derrick damage recognition network under the combined effect of uniform and moving loads.The research results show that the identification of derrick damage can be completed by two measurement points under the joint action of moving load and uniform load.With Caofeidian Bridge as the background,the neural network is used to learn the relationship between the increase of deflection increment of the main girder and the damage position and degree of the suspender system,and the health monitoring of the suspender system of the arch bridge is realized.This paper focus on the single derrick damage,double derrick damage two types of damage condition,combination with the engineering practice of sampling methods,the identification network under uniform load is optimized to take into account the fact that the location and extent of damage to the arch bridge derrick system can still be accurately monitored by deflection increments when the combined effect of temperature or moving loads is applied under uniform load.