把神经网络应用于丝杠磨削过程的建模与控制

提出了利用两个人工神经网络对丝杠的磨削过程进行建模与预测控制的思想.其中,网络1用于复映传动链、热变形和力变形等误差源与工件螺距误差的关系,即建模;网络2根据网络1的输出和工件螺距误差的仿真值而预报输出下一采样周期的综合补偿控制量.通过一系列试验研究,证明此控制策略能减少工件螺距误差80%以上,有效提高了试件丝杠的磨削精度.

Model establishment and control of the grinding process of a leadscrew by artificial neural network

A strategy for model establishment and predictive control of the grinding process of a leadscrew by two artificial neural networks is proposed. The first ANN is used to model the relationship of thermal deformation and force deformation error with the pitch error of the workpiece ,and the second ANN outputs the final modified control value according to the output of the first ANN and the theoretical control value, and the modified control value can be applied for compensation control at the beginning of the next sampling period. The experimental results show that the pitch error of the workpiece is decreased by more than 80%, thus significantly improving the grinding precision of the leadscrew.