Injury Prediction of Nonstandard Posture Occupants in Typical Intelligent Vehicle Frontal Impact

As future intelligent vehicles are gradually brought into use, the vehicle driving scenario may show a predictable and highly diversified trend, which introduces a new challenge to the design of occupant injury protection in vehicle collisions. In this regard, quantitatively analyzing the sensitivity of occupant injury responses to highly diverse occupant postures and accurately predicting the risk of occupant injury related to specific OOP (out-of-position) postures are crucial for designing the crash safety of intelligent vehicles. In this study, a variety of typical occupant OOP postures are considered. Numerical simulations are conducted with existing occupant restraint systems and human body numerical models to investigate the correlation between OOP posture factors (such as AEB (automatic emergency braking) level, torso left/right tilt angle, body size, and seatback tilt angle) and the predicted injury responses. An OOP occupant injury prediction model for frontal collisions is established using BP (backpropagation) neural networks based on the simulation data. Our study shows that the OOP posture factors are correlated with various injury criteria to different extents. The seatback tilt angle is significantly and positively correlated with neck injury and is also clearly related to upper leg injury. The occupant torso left/right tilt angle is remarkably negatively correlated with chest injury, especially chest compression. The occupant size mainly affects lower leg injury, particularly tibia force, while the effect of AEB is less notable than these factors. The BP neural network-based model for the prediction of OOP occupant injury showed Adequate prediction performance in terms of consistency index (higher than 0.9 for all injury criteria) and prediction accuracy (higher than 84 for all injury criteria, except for the viscous criterion VC and neck extension bending moment). This study may serve as an important reference for the design of intelligent occupant restraint systems in crash events involving intelligent vehicles.