神经网络和遗传算法在齿轮注塑成型工艺参数优化中的应用

Application of Neural Network and Genetic Algorithm in the Process Parameters Optimization of Gear Injection Molding

建立了基于神经网络和遗传算法并结合正交试验的齿轮注塑成型工艺参数优化系统。用正交试验法来设计神经网络的训练样本，人工神经网络有效的创建了翘曲预测模型；遗传算法完成了对影响薄壳塑件翘曲变形的工艺参数（模具温度、注射温度、注射压力、保压时间、保压压力和冷却时间等）的优化，并计算出了它们的优化值。

Process parameters optimization system of gear injection molding was built based on neu-ral network and genetic algorithm, and combined with the orthogonal experiment. Designed training sam- ples of neural network according to orthogonal experiment method, creating effectively warp prediction model. Based on genetic algorithm, completed the process parameters optimization which effects the deformation of thin shell plastic parts warp, as well as calculated the optimal value.