Generation of conceptual design scheme of automobile wheel based on reinforcement learning and topology optimization
In view of the fact that the current conceptual design of automobile wheel hub focuses on theoretical research,there are subjectivity and limitations when verifying the validity of the theory,and the generated design scheme has few options or does not consider the engineering performance,a method of generating conceptual design scheme of automobile hub based on reinforcement learning ( RL) and topology optimization is proposed to generate multiple feasible conceptual design schemes during the design of automobile wheel hub structure. Firstly,topology optimiza-tion analysis model for automobile wheel hub is established using topology optimization theory. Secondly,the gener-ating process of conceptual design scheme combining solid isotropic material with penalization ( SIMP) and Monte Carlo algorithm is studied. Softmax algorithm is used to improve the searching ability when exploring the design space,so that the design model could generate multiple conceptual design schemes that met the engineering per-formance index. Finally,the mechanical properties of the conceptual scheme after detailed design are verified by the finite element analysis tool. An example of lightweight two-dimensional design of automobile wheel hub is taken to verify the effectiveness of the method.
万方数据
