BP NEURAL NETWORKS FOR TOOL LIFE PREDICTION IN HIGH SPEED MACHINING
Tool failure may result in a loss in surface finish and dimensional accuracy of the finished parts, or possible damage to the work piece and machine. This paper presents an approach by using neural networks to predict tool lifetime in high speed precision machining. A back propagation (BP) neural network has been applied for monitoring the cutter conditions, so that early discovery of tool wear could be detected and necessary adjustments can be made to prolong the life-cycle of the cutter tool. Experiments were carried out on a high speed CNC milling machine with 6mm ball-nose milling cutters. Eight channel signals are captured using three different sensors and sixteen features are therefore extracted from the raw data. Four features are defined as major influencing factors to cutter flank wear and used to build predictive reference models for prediction of cutter behaviors and useful life. A comparison is made in the case study on the prediction performances of the BP neural networks and multiple regression models (MRM) with the same set of experimental data. Back propagation neural network shows clear better performance over the MRM method for solving the problem of tool life prediction in high speed machining.Key Words: Neural networks, Milling machining, Prediction.
Neural networks Milling machining Prediction.
X. Li J.H. Zhou B.S. Lim S.J. Phua S. Huang C.K. Shaw
Singapore Institute of Manufacturing Technology, 71 Nanyang Drive, Singapore 638075
国际会议
北京
英文
874-881
2008-10-27（万方平台首次上网日期，不代表论文的发表时间）