The goal of the optimization on injection molding process is to ensure both the quality and efficiency under remaining equipment and materials to achieve the ideal forming process. Firstly, the objective function with quality and efficiency was confirmed, then the integrated optimization model was constructed with multi-objective decision variables of mold temperature, melt temperature, packing pressure, dwell time and cooling time. Furthermore, the sample data of model were trained by the powerful non-linear interpolation and self-learning ability of BP network. In the end, the optimal injection designing process was obtained by combining the global and multi-objection optimization ability of the improved Non-dominated Sorting Genetic AlgorithmsⅡ.