关键词: 蜗杆砂轮磨齿机, 概率神经网络, 热误差, 建模

Abstract: The radial thermal error was the key factors of thermal error which directly affected M value of gear in CNC worm wheel gear grinding machine.Therefore,a method for the radial thermal error compensation of CNC worm wheel gear grinding machine based on Probabilistic Neural Network(PNN) was put forward.The temperature variables was optimized and selected by adopting the methods of fuzzy clustering,and the thermal error compensation model were set up by PNN.The compensation model was tested with CNC worm wheel gear grinding machine.The result showed that the error of the gears M Value was reduced from 0.06 mm to 0.01 mm or less.The above research indicated that the PNN model had high accuracy for thermal error modeling of CNC worm wheel gear grinding machine.Through comparing with other model,the stability and the feasibility of this model were verified.

Key words: worm wheel gear grinding machine, probabilistic neural network, thermal error, modeling