摘要
针对数控机床热误差建模补偿的问题,提出了灰色神经网络建模补偿的新方法。首先利用机床的温度值建立了机床热误差的灰色系统预测模型,再由灰色模型预测值得到的残差建立神经网络预测模型。结合灰色系统和神经网络的优点,建立了一种新的灰色系统和BP神经网络组合热误差预测模型。最后以实测数据建模说明了灰色神经网络模型预测效果明显优于各单项模型,方法优异的预测性能对于具有复杂成分的动态数据序列的机床热误差建模也适用。
A new method of gray neural network for thermal error compensation modeling is put forward. Based on temperature of NC machine tools, the grey system model used in thermal error forecasting is proposed, and builds neural network forecasting model by use of data series from grey system predicted value decrease original value. Synthesizing the advantages of grey system theory and neural network, a new prediction model named GNN combined forecasting model is applied to predict the trend of thermal errors in machine tools. Finally, efficiency of GNN forecasting model is demonstrated by an example and showes higher prediction accuracy than the single model. With excellence performance on prediction, this method is of significance for setting up thermal error modeling and being suitable for modeling of dynamic data sequence with complicated components.
出处
《机械设计与制造》
北大核心
2014年第7期236-239,共4页
Machinery Design & Manufacture
基金
江苏省镇江市科技支撑计划(GY2012028)
江苏省(丹徒辛丰)轴承产生公共技术服务中心(BM2011105)