Abstract: The widely used contact modeling methods are mainly based on traditional physical contact models, such as Hertz model and nonlinear contact models. However, their simulation results are not accurate enough, which makes them more suitable for the trend analysis of contact phenomena. In this paper, a hybrid contact modeling method, based on the traditional physical model and the data-driven error model, is proposed to accurately simulate the contact phenomenon. The physical contact model simulates the known contact factors in the contact phenomenon, and the data-driven error model based on the radial basis function neural network model describes the errors between the contact phenomenon and the simulation result of the physical contact model. A bouncing ball contact test is proposed to verify the hybrid contact modeling method, and the test result shows the feasibility of the hybrid contact modeling method and the accuracy of the hybrid contact model.

Key words: contact dynamics, contact modeling method, data-driven model, radial basis function neural network model, improved PSO algorithm based on genetic operator