Parameter optimization of CF/PEEK composites and 7075 aluminum alloy by resistance implant welding based on machine learning

In this paper, the carbon fiber/polyetheretherketone (CF/PEEK) composites and 7075 aluminum alloy were taken as research objects, and the resistance implant welding process parameters were investigated. Firstly, the resistance implant welding was carried out by using an orthogonal design of experiments. The welding current, welding time, welding pressure, and clamping distance were identified as four main process parameters affecting the single lap shear strength of welded joints. Then, a BP neural network and genetic algorithm were employed to optimize the welding process parameter combinations. The results indicate that the BP neural network can describe the complex mapping relationship between the single lap shear strength of welded joints and welding process parameters, and the predicted mean relative error is 3.62%. The welding current of 15.05 A, welding time of 155 s, welding pressure of 0.99 MPa, and clamping distance of 2.34 mm are the optimal resistance implant welding parameters, and the maximum single lap shear strength of the welded joint is 19.26 MPa, which is 31.2% higher than that before optimization.