【摘要】 主轴热误差是影响机床精度的主要因素,建立准确的主轴热误差模型是进行机床误差补偿的关键。研究了温度测点优化和神经网络建模的方法,给出了粒子群优化灰色神经网络建模的流程。开展了主轴热误差热特性试验,得到了主轴热变形随主轴转速的变化规律。基于粒子群优化灰色神经网络建立了主轴轴向伸长和俯仰角热误差模型,并与灰色神经网络和BP网络的预测性能进行了对比,结果表明该模型可有效提高网络模型的收敛性和预测精度。更多还原

【Abstract】 Spindle thermal error is the main influnce factor of machine tool accuracy. The key step of machine tool error compensation is to establish accurate spindle thermal error model. The temperature measure point optimization and neural network modeling were investigated,and the procedure of modeling based on neural network evolved by particle swarm optimization was proposed.The thermal characteristic experiments were carried out and the changes of spindle thermal deformation with rotating speed were obtained. The models of axial error and pitch error were built based on neural network evolved by particle swarm optimization. Comparing with grey neural network and BP neural network,the proposed method can effectively improve the model convergence and prediction accuracy.更多还原