An intelligent method for accident reconstruction involving car and e-bike coupling automatic simulation and...

Area of Science:

• Traffic Safety
• Computational Mechanics
• Accident Reconstruction

Background:

• Car-electric bicycle (e-bike) accidents are a growing concern due to increased e-bike usage and high rider casualty rates.
• Manual accident reconstruction is subjective, time-consuming, and limited by parameter combinations.

Purpose of the Study:

• To develop an intelligent, accurate, and efficient method for reconstructing car-e-bike accidents.
• To automate the accident reconstruction process using computational tools and optimization algorithms.

Main Methods:

• Developed an automatic operation framework integrating the MADYMO program with four multi-objective optimization algorithms (NSGA-Ⅱ, NCGA, AMGA, MOPS).
• Utilized 12 design variables, 5 objective functions, and 3 constraints for optimization.
• Reconstructed a real e-bike accident using surveillance video and validated the framework by comparing simulation results with real-world data.

Main Results:

• The intelligent method achieved optimization in approximately 24 hours with 480 automatic operations.
• NSGA-Ⅱ demonstrated the best performance, with average objective errors below 5%.
• Simulations showed good consistency with surveillance video data for rider kinematics, velocities, and head injury outcomes compared to medical reports.

Conclusions:

• The proposed intelligent method is valid for efficient and accurate car-e-bike accident reconstruction, outperforming subjective manual methods.
• This approach, combining automatic simulation and multi-objective optimization, can be applied to other accident types.
• Identified the significant effects of initial variables, providing recommendations for future reconstructions.