基于人工神经网络具有良好的非线性逼近特性,以先期人工经验为依据,利用正交试验结果作为神经网络的训练样本,建立改进的Vogl快速BP网络模型,并通过样本检测模型的预测准确度,在正交试验最佳组合处采用小步长搜索方法,结合神经网络模型进一步优化工艺参数.结果表明:结合正交试验,神经网络和先期人工经验进行球面磨削工艺参数优化可以缩短参数优化时间,提高工艺设计效率,改善表面加工质量. Based on the advanced artificial experience, the artificial neural network (ANN) is used as a training sample of neural network to establish an improved Vogl fast BP neural network model. The model is used to test the predictive accuracy of the model, The optimal combination of orthogonal test, small-step search method and neural network model were used to optimize the process parameters.The results showed that optimizing the parameters of spherical grinding process by orthogonal test, neural network and prior artificial experience can shorten the time of parameter optimization , Improve process design efficiency and improve the quality of surface processing.