Abstract: To improve the efficiency and objectivity of accident judgments, an automatic optimization method of vehicle-pedestrian accident reconstruction based on numerical simulation is proposed. In this method, the accident process is reconstructed based on full scale dummy multi-body simulation, and its initial conditions are automatically adjusted to optimum solution which can generate the best match between simulation and real accident data. For several types of pedestrian accident, the corresponding optimization model is presented in which the impact speed and/or the braking delay are regarded as the design variables and the displacement of vehicle and pedestrian is used to construct the objective function with the constraints of pedestrian injuries. Because sequential linear programming is used as the optimization algorithm and the optimization is based on rest position as well as pedestrian injury, the above mentioned method is suitable for the nonlinearity of pedestrian accident reconstruction. Using the commercial software, the method is applied to a real-world pedestrian accident in order to prove its feasibility. The simulation is compared with the autopsy report in terms of several pedestrian injuries. Also, the global response surface of the objective function constructed by a full factorial experimental design is used in the evaluation.

Key words: Accident reconstruction, Multi-body simulation, Optimization, Vehicle-pedestrian collision