CAJ | 학술논문

为了提高数控机床热误差模型的预测精度,以某型号立式加工中心为实验对象,采用模糊聚类与灰色综合关联度相结合的方法对机床测温点进行优化,将测温点从8个减少到3个。利用遗传算法( GA)优化的Elman神经网络建立了主轴热漂移误差预测模型,通过实例比较了 GA-Elman神经网络模型与普通Elman 神经网络模型的预测效果。结果表明,与普通Elman神经网络所建的预测模型相比,GA-Elman神经网络模型对主轴轴向热漂移误差的预测精度较高,残差较小,网络的泛化能力较好。
위료제고수공궤상열오차모형적예측정도,이모형호입식가공중심위실험대상,채용모호취류여회색종합관련도상결합적방법대궤상측온점진행우화,장측온점종8개감소도3개。이용유전산법( GA)우화적Elman신경망락건립료주축열표이오차예측모형,통과실례비교료 GA-Elman신경망락모형여보통Elman 신경망락모형적예측효과。결과표명,여보통Elman신경망락소건적예측모형상비,GA-Elman신경망락모형대주축축향열표이오차적예측정도교고,잔차교소,망락적범화능력교호。
In order to improve the prediction accuracy of CNC machine thermal error model, a vertical ma-chining center was taken as a research object. The temperature measuring points of machine are optimized by using a combined method of fuzzy clustering and grey comprehensive relationship degree. The tempera-ture measuring points were reduced from 8 to 3. The prediction model of spindle thermal drift error was es-tablished based on genetic algorithm optimization Elman neural network. The predictive effect of GA-Elman neural network model and the common Elman neural network models were compared through the example. Compared with the prediction model built by the ordinary Elman neural network, the results show that the Elman neural network has higher fitting accuracy, smaller residual error and better generalization capacity on spindle axial thermal drift error.