Prediction and detection of cutting tool failure by modified group method of data handling |
| |
| Affiliation: | 1. Toyama Pref. College of Technology, Toyama, Japan;2. Toyota Technological Institute, Nagoya, Japan;1. Department of Mechanical Engineering, College of Engineering Pune, India;2. Centre for Automation, Vellore Institute of Technology, Chennai, India;1. Department of Architecture Engineering, University of Sistan and Baluchestan, Zahedan, Iran;2. Civil Engineering Department, University of Sistan and Baluchestan, Zahedan, Iran;1. Department of Automation, Xiamen University, Xiamen 361000, PR China;2. School of Computer Science, Minnan Normal University, Zhangzhou 363000, PR China |
| |
| Abstract: | This paper describes a method of predicting and detecting cutting tool failure in a process computer by means of a statistical model formed by the group method of data handling (GMDH).An algorithm for modifying a process model, which has been formed beforehand by GMDH, using the newest real process data is derived at first.The first application of the algorithm is directed to the prediction of cutting tool wear. It has been found that the relative prediction errors fall within ±10% at more than 90% of predicting points.The second application of the algorithm is directed to the detection of cutting tool failure by chipping. The dynamic model of the cutting torque signal is formulated by GMDH and continuously renewed using time series process data. The difference between the estimated process output and the real process data is always monitored and becomes remarkably large when the tool failure by chipping occurs. |
| |
| Keywords: | |
| 本文献已被 ScienceDirect 等数据库收录！ |
|
