基于神经网络理论的数控插补原理与方法

Numerical Control Interpolation Principle and Method Based Neural Network Model

在分析神经网络模型模拟非线性函数的学习训练过程及曲线曲面的数控插补原理的基础上，提出一种基于三层ＢＰ网络的任意空间曲线插科方法，并以平面内三次曲线为例，进行仿真计算。结果表明，选取合适的插补周期以及足够的隐层节点数，神经网络可以精确地模拟曲线。

This paper analyses the learning and training process of neural network which is used to simulate non-linear function and the numerical control interplation principle of curve and curved surface. Then, a method by which a space curve can be interpolated with a 3-layer BP model is developed also, the simulation calculation for an example of cubic curve in plane is given. Finally,the simulation results show that the neural network with a suitable interpolation cycle and enough hidden layer nodes can accurately simulate curves.