High-accurate cutting forces estimation by machine learning with voice coil motor-driven fast tool servo for micro/nano cutting

Abstract

In a machining process, cutting force is one of the most important machining parameters which can directly reflect the state of tool and quality of the machined workpiece. The cutting force in micro/nano cutting is required to be measured with a high accuracy of a few millinewtons. However, traditional cutting force indirect estimation methods are difficult to meet the high-accurate measurement requirement. In this paper, a reliable and high-accurate cutting force estimating method based on machine learning for a fast tool servo driven by Lorentz force without using additional sensors is proposed. A concept of virtual forces, which could enable obtaining sufficient type of forces than that by actual cutting, is developed to increase the data training accuracy. In addition, an effective data filtering method combining the fast Fourier transform and the inverse fast Fourier transform is adopted to further improve the estimation accuracy. Cutting force estimated by the proposed method is compared with that estimated by traditional disturbance observer and Kalman filter, which demonstrates a much better estimation accuracy. The experiment results show that the estimated cutting force is in good consistent with the best commercial dynamometer. Even the details of small cutting force fluctuation on the order of 50 mN can be clearly identified. The experimental results demonstrated that the cutting force estimation method for micro/nano machining process could have a high measurement resolution of 10 mN and relatively measurement error as small as 0.4%.