Self Adaptive Optimization Algorithm Based On Multiple Surrogates And Its Application In Shaft Clinching Of Wheel Hub Bearing Units

With the improvement and development of computer technology, manufacturing engineering and theory of numerical simulation, the optimization design problems of different modern engineering structure are becoming increasingly comprehensive and complicated. According to the existing problems when selecting the surrogate model beforehand mainly by the past experience in the optimal designs based on the surrogate model, in order to obtain the respective best surrogate model for each object variable and bounded variable in different optimization designs of engineering structure, the objective of this paper is to present a universal and effective self adaptive optimization algorithm based on the multiple surrogates. Moreover, the self adaptive optimization algorithm in this paper is applied to the shape optimization design of the rivet head in the shaft clinching assembly process optimization of the automobile wheel hub bearing unit. Based on the above research ideas, the research contents in this paper mainly include the following aspects.(1) According to the theory of the surrogate model technique based on the approximation method, the basic idea of the self adaptive optimization algorithm based on the multiple surrogates is proposed, and multiple surrogates consist of six kinds of surrogate models, such as the polynomial response surface, radial basis functions, Kriging, multivariate adaptive regression splines, support vector regression and improved optimal weighted surrogate. Meanwhile, using MATLAB mathematical software as main programming tool, the whole process and some key technologies of the self adaptive optimization algorithm are elaborated.(2) The self adaptive optimization algorithm is verified and analyzed by four representative mathematic optimization models. According to the situation of five surrogate model selection rounds in each test example, it shows that the ultimate approximate optimal solutions obtained by the self adaptive optimization algorithm are all close to the corresponding globally optimal solutions in theory, and the relative errors are all within 5%. Therefore, the feasibility and accuracy of the self adaptive optimization algorithm are verified effectively.(3) The self adaptive optimization algorithm is applied to the shaft clinching process optimization of automobile wheel hub bearing unit. The effective and reliable finite element model is an important foundation of the self adaptive optimization algorithm, so the modeling parameters of finite element model about the shaft clinching process are determined by the combination of the theoretical derivation and the experimental testing, and the finite element model is built to simulate the shaft clinching process of automobile wheel hub bearing unit accurately. Moreover, in order to validate the finite element model of the shaft clinching process effectively, the simulated axial riveting load-time curve and the ultimate deformed shape of the hub shaft are compared with the experimental ones.(4) Finally, the optimal design of the rivet head forming surface is realized by the combination of the self adaptive optimization algorithm and the verified finite element model about the shaft clinching process. The result shows that the energy consumption in the shaft clinching process of the wheel hub bearing unit decreases by 12.5% effectively. In addition, the product quality and some performance indicators of the wheel hub bearing unit are improved in large extent.