神经网络优化技术在磨床主轴优化中的应用

OPTIMIZATION OF THE SPINDLE SYSTEM OF SURFACE GRINDING MACHINE BY NEUROCOMPUTING

利用拉格朗日乘子法将约束优化设计问题转化为求解拉格朗日乘子法极值的问题,讨论拉格朗日乘子法极值存在的条件,基于人工神经网络非线性优化计算的优点,构造用于拉格朗日乘子法的人工神经网络学习规则和网络结构。优化计算平面磨床主轴系统的结构参数,讨论优化结果与KuhnTucker准则优化结果的一致性。

Example of the spindle system of surface grinding machine by neurocomputing is given. It is transposed that constrained optimization problem to solving Lagrange multiplier's extreme value by Lagrange multiplier's method. It is introduced Lagrangian's stationarity conditions. The artificial neural network architecture and learning rules of Lagrange multiplier's method is established. Optimum result of the spindle system of surface grinding machine is compared with that used Kuhn-Tucker criteria.