Optimization of fixture layouts of glass laser optics using multiple kernel regression

Abstract

We aim to build an integrated fixturing model to describe the structural properties and thermal properties of the support frame of glass laser optics. Therefore, (a) a near global optimal set of clamps can be computed to minimize the surface shape error of the glass laser optic based on the proposed model, and (b) a desired surface shape error can be obtained by adjusting the clamping forces under various environmental temperatures based on the model. To construct the model, we develop a new multiple kernel learning method and call it multiple kernel support vector functional regression. The proposed method uses two layer regressions to group and order the data sources by the weights of the kernels and the factors of the layers. Because of that, the influences of the clamps and the temperature can be evaluated by grouping them into different layers.

Grasping point optimization for sheet metal part based on GSA-Kriging model in a multi-robot assembly system

“…Xing and Wang [18] proposed a two-stage layout optimization method for sheet metal parts, and optimized the layout of sheet metal parts based on radial basis function and particle swarm optimization algorithm. In order to minimize the surface shape error of glass laser components, Su et al [19] proposed a multi-kernel support vector function regression model, and constructed a two-layer support vector machine regression model of device deformation response with respect to ambient temperature and clamping force. Subsequently, the branch and bound algorithm were used to optimize the clamping force.…”