Significance of CAE simulation thus is increasing because of its ability to predict the failure faster, also lot of design combinations can be evaluated with this before physical testing. Frame stiffness of side doors is one of the major criteria of a vehicle closure system. In most cases, designers around the globe will be designing same or very similar side door frame structures recurrently. In addition, in the current growing trend having an optimized side door frame design in quick time is very challenging. In this investigation, a new artificial intelligence (AI) approach was demonstrated to design and optimize frame reinforcement based on machine learning, which has been successful in many fields owing to its ability to process big data, can be used in structural design and optimization. This deep learning-based model is able to achieve accurate predictions of nonlinear structure-parameters relationships using deep neural networks. The optimized designs with optimization objectives as deflection is obtained efficiently and precisely using Bayesian Optimization algorithm. Deep learned computational results were validated by the experimental results. Furthermore, the developed deep neural networks show how various design sections of door frame structures behave and how it can be used as a reference for future door design through deep generative model techniques.
Keywords: Door Frame Reinforcement, Deflection, Deep neural networks, Bayesian Optimization algorithm