Prediction of Slip and Torsion Performance of Right-Angle Fasteners Based on Machine Learning Methods

Abstract

Aiming at the issue of large CPU costs and low calculation accuracy in the design of right-angle fasteners in scaffolding structures, prediction models of fastener anti-slip performance and torsion performance based on machine learning are proposed. A three-dimensional solid model of right-angle fasteners is established using the finite element method, the effectiveness of the numerical simulation method is verified through test results, and the comprehensive influence of various design parameters on the performance of fasteners is revealed by the parametric analysis method. The database is established by combining the test and numerical simulation results, and the fastener stiffness prediction models are proposed based on random forest (RF), support vector machine (SVM) and K-most proximity algorithm (K-NN), respectively. The expressions for the measured point displacement of the anti-slip model and the stiffness prediction of the torsion model are proposed in combination with genetic expression programming. The results indicate that SVM and GEP can predict the displacement and torsional stiffness of right-angle fasteners more accurately, which is important for guiding the safety design of fasteners in engineering scaffolding structures.

Keywords

• rectangular fasteners
• anti-slip performance
• torsional performance
• parameter analysis
• machine learning