Research And Implementation Of Multi-channel Vehicle Weigh-In-motion System

In the last ten years,with the development of national infrastructure,highway construction has gradually increased.The transportation industry is booming,which brings about the problem of highway management,mainly focusing on overloading.Based on the algorithm performance and working mode of WIM(Weigh-In-Motion)system,this thesis aims to improve and innovate the traditional WIM system in the market,improve the accuracy of the system and strengthen the real-time performance of the system.In the case of multiple lanes,all lane vehicles are weighed orderly and reliable.A multi-channel vehicle WIM system is designed to ensure real-time,stability and reliability.The system needs to have the ability to transmit the weighing signal in real time,to calculate the signal accurately,and to match the vehicle information accurately.By reading and summarizing a large number of documents and combining with the requirements of the system,this thesis puts forward the main research direction and specific work contents,and explore the core units of the WIM system in the following four aspects:(1)For WIM system development and future functional expansion and upgrading,this thesis will adopt a CPU&FPGA platform architecture.The development and testing of each function module of the core unit in the multi-channel WIM system is completed,which ensures the smooth communication link.The real-time data transmission module based on discrete architecture is designed,and the test experiment of data transfer is completed.The module coordinates the data collected and stored by the FPGA and moves the data between the ARM cores at high speed.For the middle layer of hardware and software,the application program interface(API)is designed to improve the development efficiency and test its applicability.(2)To solve the noise problem of multi-channel WIM system,the collected tested by Matlab.The traditional noise suppression method of WIM system and the denoising method based on wavelet transform are used to compare the experiments,and then according to the characteristics of singular points based on wavelet transform,the starting and ending points of weighing waveform signal are found to improve the calculation accuracy of weighing waveform.(3)To solve the problem of the calibration of WIM system is not flexible in engineering,the calibration function is obtained by studying the factors that cause the deviation by studying the guidance of pavement excitation.This function eliminates the constant term in the traditional weighing algorithm to improve the flexibility and reliability of system calibration.(4)To solve the problem of cross-lane vehicles-separate error under multi-channel,the cross-lane vehicles-separate is classified by logic,and the problem is abstracted into statistical classification problem by establishing mathematical model.A simulation data set is established to explore the classification method of machine learning based on SVM,and to verify the applicability and rationality of the data set.