关键词: 布置优化, 柔性薄板装配, 不等间距定形焊点, 支持向量回归机, 粒子群优化算法, 代理模型

Abstract: Aiming at the problems that the sheet metal assembly variation was influenced by shaping welding spot layout and the welding spot equal pitch layout was not be appropriate in all cases of sheet metal assemble,an optimization method of shaping welding spot unequal pitch layout for reducing assemble variation based on Support Vector Regression (SVR) and Particle Swarm Optimization (PSO) was proposed.The sheet metal assembly variation model which described the relationship between assembly variations and variation sources was established through the analysis of sheet metal assembly process.Two pitches of welding spot were taken as design parameters to obtain the sample schema of welding spot layout in design space of layout by optimizing Latin hypercube design,and the sample response was obtained by assembly variation model.PSO-SVR surrogate model was established to predict the best shaping welding spot unequal pitch layout which minimized the assembly variation of sheet metal.A simple sheet metal assembly was used to indicate the effectiveness and superiority of the proposed method by comparing with two surrogate models that were GA-BP neural network and response surface method,and its practical was indicated by an example of auto-body sheet metal assembly.

Key words: layout optimization, sheet metal assembly, shaping welding spot unequal pitch, support vector regression, particle swarm optimization, surrogate model