Enhancement of Accuracy in Sensorless Cutting-Force Estimation by Mutual Compensation of Multi-integrated Cutting-Force Observers

Abstract

Techniques without external sensors for cutting-force estimation based on the internal information of machine tools are attracting attention from the viewpoints of cost and sustainability. It is known that using a cutting-force observer, which can estimate the cutting force from encoder information and current reference value, is effective. To date, various cutting-force observers have been proposed. However, each cutting-force observer has different characteristics, such as the bandwidth and direction, which can be estimated with high accuracy. Therefore, the variation in the machining conditions causes a decrease in accuracy when estimating the cutting force. In this study, the output data from multiple cutting-force observers were integrated and mutually compensated for by building a linear regression model and applying the least-squares method to determine the calibration factors. The validity of the proposed method was evaluated using end-milling tests. The experimental results showed that the estimation accuracy of the cutting force using the proposed method can be greatly improved in the feed and cross-feed directions compared with the conventional methods.

Keywords: Process monitoring; Linear regression model; Cutting-force observer