Study On The Key Technologies Of Pedestrian-Vehicle Traffic Accident Reconstruction

With the increasing number of motor vehicles and the gradual standardization of handling procedures for traffic accidents,the demand for professional accident analysis is also significantly increased.The main task of accident reconstruction,which is both important means and important content for accident analysis,is to explain and reproduce the accident process and impact states with certain methods based on the on-spot information.Pedestrian-vehicle accident is one common type of accidents on the road,but until now there is no methodological system for pedestrian accident reconstruction.The main content of this dissertation is to focus on the key technologies for pedestrian accident reconstruction.The dissertation generally proposes and construts the methodological system for pedestrian accident reconstruction including five key issues: in-depth accident investigation method,impact modeling method for accident simulation,pedestrian projection kinematics,uncertainty analysis method and optimization menthod for accident reconstruction.In chapter 2,a framework for in-depth pedestrian accident investigation is proposed based on partipation in the construction of NAIS project(National Automobile Accident Investigation System)since 2011.An UAV(Unmanned Aerial Vehicle)and image correction program for UAV are designed for investigation of accident scenes fastly and totally,the method to rebuild accident scene from photos is also researched for extracting valuable information after removal of accident scenes.Besides,the methods to rebuild road alignments,to test adhesion coefficient and friction coefficient,are discussed.Four methods for analyzing impact speed are provided including surveillance-video-based,braking-distance-based,kinematics-based,and EDR-based methods.In the end of Chapter 2,181 real-world pedestrian accidents from NAIS in-depth data(2011~2016)are analyzed in the view of characteristics and mechanism of fatal pedestrian-involved accidents.In chapter 3,two kinds of accident simulation methods for accident reconstruction are researched,including rigid-body-based and multi-body-based method.The basic principles,core algorithms,and modeling approaches for the two methods are also provided and discussed followed by a real-world application with simulation platform ARAS and PC-Crash.In chapter 4,three types of vehicle are focused,including low-long-fronted vehicle,high-long-fronted vehicle and minivan,which are high-frequently involved in pedestrian accidents on China's roads.With a large number of simulation experiments,the pedestrian projection kinematics are discussed from the aspects of launch angle,apogee height,first landing point distance,and throw distance in typical impacts with the three types of vehicles with impact speed from 20 km/h to 110 km/h.Three kinds of projection(wrap,bump,and arch)are founded and explained followed by new throw distance models with power function;the influences of other impact factors are also discussed including impact position,vehicle parameters,pedestrian speed and posture.In the end of Chapter 4,295 real-world accidents(including 91 real-world accidents with surveillance video)from NAIS database,and 108 real-world accidents from CASR database are provided to verify the new models proposed by this dissertation.In chapter 5,several uncertainty analysis methods for pedestrian accident reconstruction are researched based on definition for the issue,including upper and lower bound method,finite difference method,uncertainty evaluation algorithm and Monte Carlo method.A new method(named “Implicit Simulation –Monte Carlo method”,ISMC)for uncertainty analysis under accident simulation is proposed,the algorithm and real-world application are also provided to verify it.In the end,a calculation example is provided to compare the application scope,calculation efficiency and accuracy for the five methods.In the last chapter,three manual optimization methods for computer simulation-based pedestrian accident reconstruction are proposed,including golden section method,compound form method,random test method.Basic principles of the three methods is elaborated,and iterative steps and processes of the algorithm are also developed.Real-world applications are also provided to verify them.