CAJ | 학술논문

为了减小主轴热误差影响，提高机床的加工精度，提出了基于时间序列和神经网络的主轴热误差综合模型。选用VM850立式加工中心为试验对象，对主轴的关键点温度和热误差数据进行采集，并将温度数据作为神经网络的输入变量，历史热误差和温度数据作为时间序列的输入变量，通过调整权重系数建立主轴热误差的综合预测模型。将综合预测模型和传统的时间序列模型以及神经网络模型进行对比试验，结果表明：该综合预测模型减少了关键温度点的数量，并且获得了较高的建模精度和模型鲁棒性。
위료감소주축열오차영향，제고궤상적가공정도，제출료기우시간서렬화신경망락적주축열오차종합모형。선용VM850입식가공중심위시험대상，대주축적관건점온도화열오차수거진행채집，병장온도수거작위신경망락적수입변량，역사열오차화온도수거작위시간서렬적수입변량，통과조정권중계수건립주축열오차적종합예측모형。장종합예측모형화전통적시간서렬모형이급신경망락모형진행대비시험，결과표명：해종합예측모형감소료관건온도점적수량，병차획득료교고적건모정도화모형로봉성。
In order to reduce the impact of spindle thermal error and improve the precise of the machine tool, a novel method combining times series and neural network is proposed for thermal error modeling in machine tools. Temperature and error datum are collected by a test designed on the VM850 machining cen-ter , and the thermal error compensation model is established by using temperatures as input datum in NN and using errors as input datum in TS. The results show that this novel error compensation model performs much better than the traditional TS and NN models. It requires less temperature sensors and has a good calculate precision and robustness.