Development of FEA-ANN hybrid model for Equivalent Stress prediction of automobile structural member

Abstract

Chassis serves as a backbone by supporting the body and diverse parts of the automobile. It ought to be sufficiently rigid to endure the shock, twist, vibration and extra stresses. Then, a vital consideration in chassis design is the strength (Equivalent Stress) for sufficient bending stiffness (Deflection). The primary goal of the research is to build up an Artificial Neural Network (ANN) model for identical stress prediction. Two side members joined to a series of cross members to make the chassis frame. The number of cross members and their locations, cross-section and the sizes of the side and the cross members turn into the design variables. The chassis frame model is created in Creo 3.0 and dissected using Ansys. Since, the number of parameters and levels are more, so the probable models are too much. By changing the Parameters, using the orthogonal array the weight of the sidebar is decreased. Then, FEA is performed on those models. ANN model is prepared by using the results of FEA. For training the ANN model, the standard back-propagation algorithm is observed to be the best. A multi-layer perception network is used for non-linear mapping between the input and the output parameters. FEA-ANN hybrid model can save material used, production cost and time.