Crack Automatical Detection And Evaluation And Parameters Robust Optimization Design For Aluminum Alloy Self-piercing Riveting

As an important way for energy-saving,automobile lightweight has drawn more attention from researchers.It's a suitable way to realize the lightweight design by using the lightweight materials such as Aluminum alloy and so on.In the process of lightweight materials connection,self-piercing riveting has greater performance than traditional welding,which has been more focused in lightweight research.The quality of selfpiercing riveted is mainly affected by cracks on the rivet button and the geometrical parameters of meridian section.However,current rivet button crack detection and evaluation is a manual process,which is subjective and time-consuming,and lack of qualitative analysis of parametric effect on joint quality.Therefore,this paper developed a button crack automatical detection and evaluation method for self-piercing riveting based on image processing and machine learning.In addition,this paper also established the self-piercing riveting finite element model to realize the parametric sensitivity analysis,and multi-objective robustness optimization design was carried out for the parameters with large influence.Firstly,the forming process of self-piercing riveting and the reasons behind the crack were analyzed,and cropped and preprocessed the crack areas images as training samples,to build the crack judgment network and crack classification network.This two network was used to search the cracks in the button,according to the detected crack information,the quality of the button was evaluated.Secondly,a self-piercing riveting finite element model for AA6111 T4 and HSLA 340 dissimilar materials was established,15 parameters were selected to carry out the experiment of parameter combination design,then influence of different parameters on joint interlock,maximum bottom thickness and maximum stamping force on the button was analyzed,according to the analysis,found that 5 parameters have more influence on the joint quality including die height and radius,rivet length,tangent modulus and yield limit of the rivet.In the final,multi-objective robustness optimization design was carried out for the riveting process according to the parametric sensitivity analysis results.The rivet length,die height and radius were regarded as design variable,the tangent modulus and yield limit of the rivet were regarded as noise effects,the joint interlock,maximum bottom thickness and the stamping force on the button were set as optimizing objectives and constraints,obtained the Pareto optimal solution by establishing the self-piercing riveting approximation model,the validity of approximate model and optimal design were verified by comparing the optimum design schemes and original scheme.