Measurement of high-frequency milling forces using piezoelectric dynamometers with dynamic compensation

Abstract

Piezoelectric dynamometers are widely used to measure machining forces during milling operations. While dynamometers are precise in measuring the low-frequency content of machining forces, their electromechanical dynamics distort the high-frequency content of the forces, resulting in critical measurement errors particularly in high-speed or highly intermittent milling processes. Existing methods (e.g. Augmented Kalman Filter) that are used to remediate the high-frequency content of the forces measured by dynamometers require tuning multiple parameters based on prior knowledge of the measurement noise and accurate models of the dynamometer dynamics, which continuously change during the process as material is machined away from the workpiece. Two new methods are presented in this paper to address this issue. The first method uses regularized deconvolution to estimate machining forces from the output signal of the dynamometer. In this method, regularization does not require prior knowledge of the measurement noise or variations of the system dynamics, but it cannot be implemented in online force monitoring and control systems. The second method designs a Sliding Mode Observer (SMO) to estimate milling forces recursively at each measurement timestep. The SMO can be implemented in online force estimation, is robust against the variations of system dynamics, and requires tuning only one gain that is independent from the system dynamics or measurement noise. The presented experimental results verify the effectiveness of these methods in accurately estimating high-frequency milling forces from dynamometer measurements.