Research On Performance Enhancement Of WIM System Based On Load Cells

Vehicle overload is the main cause of the road damage. With the rapid development of economy, the miles of road in China have ranked first in the world. The increasing phenomenon of overload in freight transportation has become a serious problem. The road maintenance expense caused by overload vehicles has reached billions Yuan every year. Person's injury or bridges break down is the often-occurring raffic accidents brought by overload vehicles. The huge loss of lives and wealth arise from overload vehicles. Accurate and fast weighing the axle weight of moving vehicle is significant for the traffic management, maintenance and law enforcement. In practical weigh-in-motion (WIM) system, the vehicle under weighing is required to cross the weighing plate at a speed lower than 5 km/h to guaranteeing the weighing accuracy. Therefore, it is significant and necessary to improve weighing efficiency and weighing accuracy for WIM system.Under the support of Science and Technology Commission of Shanghai Municipality ( NO. 035115003), this paper have made a thorough and comprehensive analysis on the signal processing of WIM system based on strain gauge weighing platform. A WIM experiment platform is constructed. The characteristic of WIM signal is researched. The process method based on dynamic load separation is proposed. The presented method efficiently reduces the influence of the dynamic load on the weighing accuracy and improves the weighing efficiency and weighing accuracy. The main content and contributions are summarized as follows:1. The dissertation expatiates on the background, significance, purpose and main content for research; introduces the research status of WIM and some WIM methods; analyzes the characteristics of different WIM methods; summarizes the signal process methods of WIM method based on load cells; points out that dynamic load is the important factor influencing weighing accuracy. 2. The dissertation studies the characteristic of dynamic load. According to quarter-car mode and pavement roughness mode, constructs the dynamics equation and researchs the interaction between road and tire; analyzes the contributions of tire damping and stiffness, suspension damping and stiffness, pavement roughness, vehicle speed and load to dynamic load; points out the influence of dynamic load on weighing accuracy.3. The dissertation constructs WIM system. Designs the load cells and platforms used in WIM system, does finite element ayalysis on load cell and finds the optimal location of installing strain gauge; developes software program controlling WIM system.4. The dissertation constructs transform function of WIM system according to two-freedom weighing model, proposes an adaptive dynamic correction method to compensate WIM signal and shorts the time of signal reaching plain.5. The dissertation uses empirical mode decomposition (EMD) to process WIM signal; introduces the concept of EMD; analyzes end effect and pseudo intrinsic mode function (pseudo-IMF) of EMD and their influences on weighing accuracy; presents AR mode method to extend WIM signal series and correlation coefficient to judge pseudo-IMF. The max weighing error of axle-weight is 7.38% at the speed of 20km/h or lower.6. The dissertation presents nonlinear curve-fitting method to separate the dynamic load contained in WIM signal, constructs parameter equation on the base of WIM signal; analyzes some parameter optimum methods and choices Levenberg-Marquardt method to estimate parameters'values; presents Multiple Signal Classification (MUSIC) method to estimate the frequencies of WIM signal and reduce the number of parameters to be estimated. The max weighing error of axle-weight is 5.24% at the speed of 15km/h or lower7. The dissertation emploies BP and RBF neural networks to estimate axle weight of vehicle at high speed; introduces the concept of neural network; compares the influences of structure, parameters and inputs of networks on weighing accuracy; constructs the mapping between network inputs and axle weight; points out that performance of RBF network is better than BP network's..