Casting process optimization usually includes modifying the mold structure, regulating the pouring temperature, changing pouring time, etc., which involves the mold structure optimization, but also involves optimization of temperature field. In order to get a better process conditions, it will generally make multiple simulation test under the different processes conditions. Many parameters should be set in each simulation test. In order to make casting engineers get liberation from a lot of repetitive work, multidisciplinary design optimization technology is firstly brought into the field of casting, successfully achieved ProCAST integrated into iSIGHT. The method of sample points collected is used in Optimal Latin Hypercube Design, and the building of approximation models is based on Kriging Model. On this basis, the optimization of the objective function is applied to Multi-Island Genetic Algorithm to obtain a global optimal solution. Compared to the verification result of the corresponding simulation, the relative error of the global optimal solution is 0.47%. The result of optimization is ensure average shrinkage rate of casting a smaller value when technological yield of the casting from 54.6% of the existing production case up to 61.1%.