Multi-objective optimization of gear forging process based on adaptive surrogate meta-models

摘要

In forging industry, net shape or near net shape forging of gears h as been the subject of considerable research effort in the last few decades. So in this paper, a multi-objective optimization methodology of net shape gear forging process design h as been discussed. The study is mainly done in four parts: building parametric CAD geometry model, simulating the forging process, fitting surrogate meta-models and optimizing the process by using an advanced algorithm. In order to maximally appropriate meta-models of the real response, an adaptive meta-model based design strategy has been applied. This is a continuous process: first, build a preliminary version of the metamodels after the initial simulated calculations; second, improve the accuracy and update the meta-models by adding some new representative samplings. By using this iterative strategy, the number of the initial sample points for real numerical simulations is greatly decreased and the time for the forged gear design is significantly shortened. Finally, an optimal design for an industrial application of a 27-teeth gear forging process was introduced, which includes three optimization variables and two objective functions. A 3D FE numerical simulation model is used to realize the process and an advanced thermo-elasto-visco-plastic constitutive equation is considered to represent the m aterial behavior. The meta-model applied for this example is kriging and the optimization algorithm is NSGA-II. At last, a relatively better Pareto optimal front (POF) is go tten with gradually improving the obtained surrogate metamodels.