Tool monitoring system for end-milling process using an MEMS accelerometer, PC sound card and a personal computer

• Main Author: Catalan, Paul Maderal
• Format: text
• Language: English
• Published: Animo Repository 2006
• Subjects: Microelectromechanical systems > Design and construction; Machine-tools > Maintenance and repair; Machining > Automation; Manufacturing
• Online Access: https://animorepository.dlsu.edu.ph/etd_masteral/3378; https://animorepository.dlsu.edu.ph/context/etd_masteral/article/10216/viewcontent/CDTG004054_P.pdf
• Institution: De La Salle University

A micro-electromechanical system (MEMS) accelerometer, an ordinary PC sound card and a personal computer were used to build a tool monitoring system to detect impending tool failure in end-milling process. The acceleration data from the machining process were acquired by the MEMS accelerometer and the sound card, and processed using Matlab in the personal computer. Using Matlab's ARYULE function to determine the autoregressive model parameters to fit the acceleration data, the modal energies of the data were isolated and plotted, and compared with the ideal tool wear curve. The energy plots of the first, second, third and fourth multiples of the tooth pass frequencies were considered in the study. A tool failure detection algorithm based on the plots was developed to monitor the tool wear and provided a means of predicting impending tool failure. Nine (9) tests were conducted to determine the applicability of the combination of the MEMS accelerometer and sound card in gathering and processing acceleration data for on-line tool monitoring. In the tests, three (3) sizes of tools were used to machine mild steel in a machining center using three cutting methods: the zig method, the follow periphery, and the follow part. The energy plots showed an increase in vibration energy as machining progresses and they showed similarity with the ideal tool wear curve. By developing and implementing a tool failure detection algorithm, the system was able to predict impending tool failure in at least two of the four energy plots monitored in the tests.