Real-time thickness compensation in mirror milling based on modified Smith predictor and disturbance observer

Abstract

Thickness of large thin-walled parts is difficult to keep constant since the time-varying disturbances during machining, such as the deformation of the workpiece and the machine tool errors. Mirror milling is an effective means to machine such parts. It reduces the deformation and acquires the wall thickness of the cutting area online via the support head arranged symmetrically with the cutter. However, the delay of thickness measurement makes the closed-loop system unstable and difficult to achieve satisfactory control of thickness. In this work, a novel online thickness error measurement and compensation method based on modified Smith predictor (MSP) and disturbance observer (DOB) is proposed for mirror milling. While previous researches have only realized open-loop thickness control by compensating for deformation, the proposed method achieves stable real-time closed-loop control and is not sensitive to the source of wall thickness error. Firstly, an online measurement system is built to obtain the thickness and local deformation of the workpiece. The measured deformation data and the geometrical pose of the machine tool are used to estimate the wall thickness to reduce the delay of thickness acquisition. Then, a DOB based on thickness estimation is constructed to observe and eliminate the external disturbances. Since there is small delay in the thickness estimation, the proposed DOB can obtain faster disturbance rejection. Furthermore, MSP-DOB compound controller is proposed, in which MSP is used to stabilize the closed-loop system by eliminating the thickness measurement delay and improve the accuracy by compensating for the internal disturbances. Simulations and cutting experiments are both conducted. The wall thickness accuracy is improved from [-0.101,0.065] (unit: mm) to [-0.065,0.048] (unit: mm) in the experiments, thus verifying the validity of the proposed method.