Artificial Neural Networks for Passive Safety Assessment

abstract

© 2022, International Association of Engineers. All rights reserved.¿Nonlinear analysis has been applied to evaluate passive safety systems. It is based on the mechanical responses of the car structure and loads generated in the occupants as well as pedestrians. These responses are evaluated to design the car structure to manage and prevent the transmission of impact energy and as a passive element to absorb it and dissipate it. Human responses are evaluated through biomechanical assessment to identify and reduce human injury. Small electric cars have been introduced to reduce pollution, and although they have an environmental advantage, the battery can explode if the structure of the car body does not manage the deformation energy well. Due to their maximum velocity, the small electric cars can be introduced in some regions without analysing their crashworthiness behaviour. In this work, it is proposed to evaluate the nonlinear response of a mechanical bump shock absorber using a neural network, to predict its behaviour as an alternative tool to perform nonlinear initial evaluation, because there is human injury at low velocities. A combination of deceleration level and its time duration is necessary to evaluate the injury at low velocities. A dynamic neural network has been used to predict the deceleration, kinetic energy and deformation responses of a mechanical bump shock absorber. The methodology can be used by original equipment manufacturers, start-ups, suppliers and companies related to mobility and micro-mobility to perform safety assessments.