Tool wear monitoring based on cointegration modelling of multisensory information

Abstract

A new method based on the cointegration theory is presented in this article to realise tool wear monitoring based on multisensory information. The main characteristic of cointegration modelling is that it can reflect the long-run equilibrium relationship of non-stationary time series, thereby avoiding the appearance of spurious regression. Monitoring of milling process was carried out by installing two dynamic force sensors in different directions. The cointegration model is constructed to describe the relationship between the tool wear value and the relevant features extracted from the sensory signal. For comparison, the root mean square error (RMSE), mean error (ME) and residual error distribution are utilised to evaluate the prediction accuracy of the cointegration model and the multiple linear regression (MLR) method. The monitoring and comparison results show that the proposed method can avoid the spurious regression effectively and predict the tool wear more accurately. This method casts new light on the accurate prediction of tool wear in the industrial environment.

Keywords:

• tool wear prediction
• cointegration modelling
• multiple linear regression
• multisensory information

Acknowledgement

This project was supported by the National Science Fund of China (51175371).