Research On Damage Identification Of Stay Cable Based On Neural Network

As one of the main force members of cable-stayed bridge,the cable will rust and fatigue under the influence of surrounding environment and long-term dynamic load,which will directly cause stiffness degradation and strength attenuation of cable-stayed,and reduce the service life of cable-stayed bridge.It seriously threatens the safe operation of cable-stayed bridge.It is of theoretical and practical significance to grasp the operation status of cable in real time and to scientifically evaluate the health status of stay cable.Based on bridge deflection monitoring,the damage identification of cable is realized by neural network method.The details are as follows:(1)The sensitivity of main girder deflection increment of cable-stayed bridge under dead load(dead weight)is analyzed.The main girder deflection increment is used as the input sample of neural network training to complete the cable damage identification.On this basis,the factors affecting the recognition accuracy of neural network are analyzed,and the neural network model is optimized.(2)The sensitivity of the deflection increment of the main girder of the cable-stayed bridge under the action of dead load and overall temperature rise and fall is analyzed.According to the variation law of girder deflection increment under temperature load,the training input samples are re selected,and a set of reasonable method for selecting temperature training samples is explored to identify the damage location and degree of stay cables.(3)Under the action of dead load and moving load,the deflection increment sensitivity of the main beam at different measuring points after stay cable damage is analyzed,and the position of sensitive measuring points is selected,and the moving load problem is transformed into concentrated load problem by using displacement reciprocity theorem.Complete stay cable damage identification.On this basis,the sensitivity of the deflection increment of the main beam after the cable damage is analyzed under the combined action of dead load,moving load and integral lifting and cooling.According to the law of the deflection increment of the main beam under the temperature load,the training input sample is re-selected.Combined with temperature training sample sampling method,the location and degree of cable damage are identified.Combined with the health monitoring system of the bridge,the damage position and degree of the stay cable can be successfully judged by the neural network method,which can be applied to the daily maintenance of the cable-stayed bridge and the early replacement of the cable-stayed bridge,so that the bridge health monitoring is more practical.