Development Of A Weigh-In-Motion System Based On Multiple Sensors

Development of economy and technological innovations have brought profound changes to the transportation system and urbanization in our country.Road,as a core part of our transport system,plays an important role in daily life and economic production.However,overloading of heavy vehicles has been a great concern for the infrastructures and traffic:it significantly reduces the service life of roads and bridges,increases the workloads and expense of maintenance and reconstructions,lowers the traffic efficiency,and raises the risk of traffic accidents.Current Weigh-In-Motion(WIM)systems are usually of high cost in installation and maintenance.Thus,this study presents a multi-sensor WIM system based on piezoelectric force sensors and vibration sensors.By the principle analysis and design,the prototype is tested by indoor machines and an outdoor full-scale loading system.After analysis of the original data,an algorithm is put forward based on multi-source data fusion.Finally,a FE model of the embedded WIM system and the pavement is set up to analyze the dynamic response under different conditions.The main contents and results are as following:(1)Based on the piezoelectric effect and pavement vibration,this research proposes a multi-sensor WIM system to measure axle load,vehicle speed and load lateral location.By design and prototyping,a distributed piezoceramic sensor has been accomplished for the proposed system,meeting the needs of data acquisition and environment.Considering the road low-frequency vibrations,two kinds of accelerometers are selected for the WIM system,followed by analysis of their principle and performance.(2)According to indoor loading tests,the designed piezoceramic sensor is of good structural stability and of reliable signal output.The sensor's output increases as the loading amplitude growing,and the linear correlation is 99.3%,showing its linearity to load.When loading frequency is of 5 Hz-33 Hz,the sensor's output is independent with the frequency.By comparing the units' output inside of the sensor,center of the load can be located.(3)By the full-scale Accelerated Pavement Tester,the effects of axle load,vehicle speed and load lateral position on the sensing system's output are separately analyzed,as well as the repeatability and stability of the system.Two methods are adopted to analyze the collected signal.For the piezoelectric signal,the peak area method outperforms the peak value method,with a better linear correlation(R2 of 91.3%).For acceleration signals,neither the peak area nor the peak value is significantly linear with the axle load.Speed is a key factor for the system,which can be controlled by introducing speed correction.The load lateral position can be deduced by piezoelectric sum peaks,peak acceleration and acceleration peak area.(4)With the BP neural network,three networks are established respectively based on acceleration data,piezoelectric data and multi-sensor data for axle regression model.The model of multi-sensor data performs best,with the minimum error and the highest correlation coefficient(99.88%),better than that of the piezoelectric data(99.74%)and that of the acceleration data(76.34%).Furthermore,multiple linear regression,support vector regression and Gaussian process regression are adopted for multi-sensor data fusion.The comparison results show that the BP neural network regression outperforms the other three regression models.(5)By the finite element method,a model of vehicle-pavement and a model of the piezoelectric sensor have been established and verified.Then,a model of the embedded sensing system and the pavement has been set up.For further application,a series of loading conditions have been simulated:four kinds of vehicle loads(65%,100%,135%and 180%),six groups of speeds(of 18 km/h-108 km/h),three grades of road unevenness(Level A,B,C).The corresponding strain,vibration and electrical responses of the embedded WIM system are collected and analyzed.According to the analysis and conclusions,the proposed multi-sensor WIM system based on piezoelectric material and vibration monitoring has solid advantages on the conventional road monitoring(such as WIM systems),which is also promising in the intelligent infrastructure applications(e.g.load location monitoring).