Two-phase data science framework for compensation of the friction force in CNC machines

ABSTRACT

This study proposes a two-phase data science framework for the friction force and parameter estimation of the hysteresis effect segment in the servo-control systems of precision machines. The first phase uses an exponential-based friction force model to identify the model parameters by an autoregressive model and Z-transform. The second phase uses symbolic regression for the residual analysis to enhance the friction force estimation. An empirical study of three types of CNC machines under different working conditions is conducted to validate the two-phase data science framework and identify the change of the machining displacement considered to be a critical factor affecting friction force. The exponential-based model successfully eliminates the circular spike error caused by the friction force in the tapping center machine. The results indicate that the proposed two-phase framework improves mean absolute error by 5.6% on average in the tapping center, 6.5% in the milling machine and 7.6% in the turning and milling center, respectively.

KEYWORDS:

• Friction force estimation
• prognostic and health management (PHM)
• symbolic regression
• compensation
• servo control system

Acknowledgments

This research was funded by the Delta Advanced Technology Research Center, and the Ministry of Science and Technology (MOST110-2221-E-002-163), Taiwan.

Disclosure statement

No potential conflict of interest was reported by the author(s)

Additional information

Funding

The work was supported by the Ministry of Science and Technology, Taiwan [MOST110-2221-E-002-163]; Delta Advanced Technology Research Center, Taiwan.