基于混合优化算法的注射模具参数优化研究-机械设计及其理论专业论文.docxVIP

II II Abstract The developing trend of CAD (computer-aided design) technique in injection mould is of integrated, intelligent, flexible, parallel, and so on. And some progresses have been made in the integrated research of CAD/CAM/CAE (computer-aided design/manufacture/engineering) techniques in injection mould at home, and all these progresses are improving the practicability of CAD technique in molding. However, those CAD applications are in the probationary period due to their weak practical application and it is impossible to meet all clients’ need with respect to user interface. The researches on the identification methods for mould structure parameter and molding process parameter were made with CAE techniques and optimization algorithm based on the Pro/ENGINEER. All these achievement were embedded into the design platform of injection mould in order to shorten the mould design cycle and improve the molding quality of plastic products. A hybrid optimization algorithm was developed and programmed based on an improved genetic algorithm IP-μGA (a modified micro genetic algorithm with the strategy of Intergeneration Projection) and a traditional optimization algorithm (NCONF) using finite-difference gradient. As we know, IP-μGA algorithm has the advantages of avoiding premature convergence and reaching optimal area quickly while the search behavior lasts for a long time. And the traditional optimization one can converge quickly, but get in locally optimal solution easily unless an appropriate initial point is given. So in order to utilize their advantages instead of weakness, an approximate point of globally optimal solution was found firstly using genetic algorithm to seek for several generations, and then this point was used as the initial point of traditional optimization algorithm. Once the initial point was given, the seeking behavior went on using NCONF. This hybrid algorithm can converge in a short time, and the globally optimal solution was finally figured