Research On Bridge Health Monitoring Method Based On Wireless Sensor Network

In the traffic network,the bridge has the function of connecting the traffic lines,is the throat of traffic,and has important strategic significance in economy,society,life and other aspects.With the increasing complexity and span of bridge construction structure,the requirements for the safety of bridge structure become higher and higher.However,during the service period of the bridge,due to the changes of internal structure,external environment erosion and other adverse factors,the bridge will appear structural aging,damage and other problems.When the structural damage accumulates to a certain extent,it will affect the use of the bridge,resulting in catastrophic accidents in the use of the bridge,resulting in huge losses.Due to the limitation of technology,most of the traditional bridge health assessment is based on manual inspection,but the manual inspection method has many problems,such as low efficiency,long inspection period,unstable inspection results and poor safety,which can not meet the maintenance and management requirements of modern bridges.Therefore,it is of great significance to carry out effective and fast research on bridge health technology to ensure the safe operation of the bridge.According to the needs of bridge health monitoring,this paper designs a wireless sensor node and remote terminal signal processing and monitoring module.Wireless sensor consists of multi-channel sensor signal amplification and acquisition,storage,data transmission and other modules,which can realize synchronous acquisition,storage and wireless transmission of characteristic signals such as bridge vibration and environmental signals such as temperature and humidity;remote node terminal processing module completes bridge signal preprocessing,feature extraction processing such as frequency and displacement,and bridge health classification.This paper discusses the method of bridge health classification modeling based on deep learning.Based on the bridge characteristics and environmental data,using the information processing method of deep learning,using the VGG16 convolution neural network of conv1 D layer to build the bridge health classification model,taking the characteristic parameters of the bridge such as frequency,displacement,temperature andhumidity as the training input of the convolution neural network,the health degree of the bridge beam as the target label of the supervision network learning,after the machine learning training,finally a bridge health diagnosis classification model is formed to monitor the health status of the bridge remotely.The preliminary experimental results show that: displacement,natural frequency,temperature,humidity as a variety of characteristics of bridge health,can better reflect the health of the bridge;using the multi classification recognizer of deep learning to diagnose the health of the bridge,the recognition rate is high,can more accurately judge the health of the bridge.After the experimental test of an old bridge section,the model established in this paper has a high accuracy of classification and recognition,which has a certain engineering significance.