A RBF Prediction Model Optimized By A Kind Of PSO Applied In The Car Spoke Stamping

As an important method of sheet metal plastic forming, sheet metal stamping is widely used in automobiles, ships, aerospace and other fields. Since most of sheet metal forming usually experience complicated elastic and plastic deforming. Problems such as wrinkling, twisted, thinning, rupture and tensile inequality appear in the process of sheet forming. It is important to make the process of sheet metal forming reasonable in order to improve the quality of the product. Therefore, researchers and engineers from many countries have paid more attentions to optimal designs on mould and the sheet metal stamping process.In the dissertation, an optimization process is put forward for a factory from Guangxi by utilizing Dynaform based on the theory of sheet metal forming and approximate model. Then the orthogonal experiment is designed to construct an optimized PSO-RBF model to predict the thinning of the spoke. At last, GA toolbox is used to find the best combination of process parameters. Through the improved process and optimization algorithm, it can dramatically reduce the thinning of car spoke in reverse-drawing forming and hole flanging, which has a practical significance for improving the efficiency of the production and qualities of products.Using the metal forming for spoke as the research object, this dissertation has done the following works:1. Drawing process adopted by factory is simulated by utilizating Dynaform. Based on " decomposition-comprehensive" method, process of reverse-drawing and profile for predrawing are optimized. At last, thinning in the reverse-drawing is much lower with the optimized profile.2. The drawing process method and numerical simulation theory are studied. At the same time, merits and demerits of PRS model, Kriging model, ANN model and the RBF model are studied. At last, root mean squared error and coefficient of determination are proposed to evaluate the approximate model’s precision.3. Taking thinning rate of the hole flanging as the evaluation index, the orthogonal experiment is designed with the process parameters such as punch velocity, the diameter of the pre-punching, the diameter of the hole flanging and die clearance, which dramatically affect the value of thinning. Then, the thinning rates for experiment schemes which are simulated by Dynaform are obtained. According to range analysis method, the degree of thinning which these factors affect and a reasonable process are obtained.4. In order to improve the efficiency of production, a PSO method whose Inertia weight which decreased in parabola way is proposed to improve the accuracy of prediction for RBF model. The fitting of nonlinear function provided by the optimized approximate model is better than the RBF approximate model. At last, the GA toolbox of MATLAB is used to optimize the process parameters and predict thinning rate.