基于粒子群算法的滚削齿面综合轮廓误差预测模型与试验研究

在配备有自主研发的数控滚齿系统的精密卧式滚齿机上，运用三因素两水平响应曲面法设计滚削试验，研究齿轮滚削加工刀具转速、轴向进给速度、背吃刀量对齿轮齿面轮廓误差的影响规律。根据试验结果，分析得出齿轮齿廓总偏差、齿轮螺旋线总偏差、齿距累计总偏差的预测模型及各工艺影响因素对齿面轮廓误差的作用显著程度，并以三个预测模型的附加权重值之和达到最小值为目标，建立可以评定多目标误差的齿面综合轮廓误差数学模型，运用粒子群优化算法对齿面综合轮廓误差数学模型进行分析优化，寻找最佳滚齿加工工艺参数。试验表明，采用粒子群优化算法对响应曲面法建立的齿面综合轮廓误差数学模型进行优化可以作为滚削加工前的工艺参数选取方案。

A Prediction Model and Experimental Study of Gear Hobbing Profile Errors Based on Particle Swarm Optimization

A series of 3 factors and 2 levels response surface method experiments were carried out to research the influence rules of processing parameters including the spindle speed, the feed rate in Z direction, and the hobbing depth on the gears' tooth profile errors on the precision horizontal gear hobbing machine equipped with home-made CNC system. The mathematical models of the total profile deviation, the total helix deviation and the total cumulative pitch deviation were built respectively and the influence degrees of the three gear profile errors were finished based on the statical experimental data analyses. A more appropriate model for gear profile errors was established based on the weighting factor of each deviation. A method of using particle swarm optimization to improve the gear hobbing parameters was applied based on the final mathematical model. The results show that it is feasible to preselect the hobbing parameters based on the particle swarm optimization of the gear profile errors model established by response surface method.