基于人工神经网络的转台框架结构动力修正

为解决转台框架设计中结构质量不均匀问题，利用人工神经网络的非线性映射能力对转台框架结构进行了动力修正。首先建立了转台框架振动频率与其结构参数的非线性神经网络映射关系，然后利用训练好的神经网络修正框架的结构动力模型，对网络映射的结构参数再次建模进行有限元分析计算，结果表明，该方法计算简单、使用方便，达到了转台框架各方向振动频率基本一致的结构动力修正的目的。

Dynamic modification of table structure based on artificial neural network

In order to solve the problem of inhomogeneous mass in rotary table structure design, dynamic modification was made with the rotary table structure using the non-linear mapping capability of artificial neural network. First non-linear neural network mapping was establish between the vibration frequency of table frame and the structure paremeters; then the structure dynamical model of table frame was updated using the neural network trained. The model built using structure parameters mapped by neural network was analysed using finite element method. Results indicate that uniform vibration frequency can be achieved in every direction.