Optimal linear combination of neural networks modeling for machine tool thermally induced error

摘要：

In recent years,neural network (NN) modeling methods with different architectures and training strategies are widely used in machine tool thermal error compensation field,but there is typically trial-and-error procedure involved in the modeling procedure to choose the “best network meanwhile discarding all the rest.This paper introduces a new modeling method by combining the trained NNs,namely the optimal linear combination of neural networks (OLCNN),which can integrate the useful knowledge of the component NNs and thus improve model accuracy and generalization ability (i.e.model robustness).Algorithms for selecting the component networks are presented after some associated critical issues are discussed.Real cutting experiments are conducted on a CNC turning machine to validate the effectiveness of the method.Analyses of different modeling conditions indicate that the best situation for using OLCNN is when the component networks are merely moderately trained.The modeling results show that significant improvement in the model accuracy and robustness is achieved comparing with the single best NN and simple average of NNs while the modelling effort is mostly much less than finding the “best network.It is also applicable to hybrid different modeling methods other than NN.

关键词： neural network modeling thermal error machine tool optimal linear combination.

作者: Jia-yu Yan Jian-guo Yang

作者单位: School of Mechanical Engineering,Shanghai Jiao Tong University,Shanghai,China

会议类型: 国际会议

会议名称: International Conference on Modelling,Identification and Control(模拟、鉴定、控制国际会议)

会议地点: 上海

会议语种:英文

在线出版日期: 2008-06-29（万方平台首次上网日期，不代表论文的发表时间）