Metamodeling of Numerical Simulations for Optimization of Hairpin Bending Processes

Abstract

As a result of the progressive electrification of mobility solutions, manufacturing technologies for distributed stator windings need to be improved in order to meet the automotive demand for productivity and quality. In this context, the hairpin technology already provides advantages regarding the possible level of automation and productivity in comparison to conventional winding technologies, but still shows technological weaknesses concerning process reliability. Against this background, approaches for optimized process control based on physical process models have a high potential to minimize waste in production resulting from variations in wire properties or an inappropriate machine parametrization. However, the accuracy of analytical modeling techniques is limited due to complex interactions of machine kinematics, tooling and material properties in rectangular wire bending processes. Hence, the calculation effort of more precise finite element simulations needs to be reduced in order to enable model-based process control and economic tool design. Therefore, conventional and AI-based methodologies for meta modeling of numerical process simulations are introduced and compared in this paper. Furthermore, opportunities of optimized process control of hairpin shaping processes based on efficient but also accurate metamodels are discussed.