Prediction of the mechanical behavior of steel-aluminum flow drill screw joints using artificial neural network
Qiaoying Zhou,Zhong Yang,Chengtai Hu,Junfeng Xing,Congchang Xu,Zhigang Xue,Chao Ma,Luoxing Li
DOI: https://doi.org/10.1007/s00170-023-12563-y
IF: 3.563
2023-11-09
The International Journal of Advanced Manufacturing Technology
Abstract:The flow drill screw (FDS) process has been widely used for joining plates and profiles in automotive body structures owing to its advantage of one-sided accessibility. A challenge in practical applications is how to quickly and accurately obtain the mechanical behavior of steel-aluminum FDS joints with different plate thicknesses and loading conditions. This study proposes a method that integrates experimental tests, numerical simulations, and artificial neural network (ANN) models to predict the mechanical behavior of joints. An experimentally validated joint numerical model was first used to generate the training and testing datasets of the ANN model. Based on the obtained datasets, two types of ANN models were set up. The classification model was employed to classify the failure modes of the joints, and the regression model was used to predict the peak force, failure displacement, and force–displacement curves. In the regression model, three data processing approaches were employed to improve the prediction accuracy of the curves. The best performance metrics of the model were achieved when the force–displacement curves in four failure modes were predicted independently. The feasibility of applying the well-trained ANN model to untrained datasets was successfully demonstrated. The failure modes of four untrained cases were correctly classified. The predicted force–displacement curves were in good agreement with the simulations. The average errors between the simulated and predicted peak force and failure displacement were 2.52% and 2.44%, respectively.
engineering, manufacturing,automation & control systems