Auto-tuning of feedforward friction compensation based on the gradient method

Abstract

igh performance position control of drives with friction is one of t h e requirements in machine-tool applications. A gradient method based algorithm is proposed in this article for t h e auto-tuning of feedforward friction compensation. T h e method presented guarantees robust stability for all states of the system which is of great importance for industrial applications. Theoretical considerations a r e illustrated with experimental d a t a collected from t h e drive of a vertical axis used for electro-discharge machining.

Feed-forward friction observer (FFFO) for high-dynamic motion control

“…Since the friction appears as a damping factor of the motion system a positive control error means an undercompensated friction disturbance. For this reason λ (·) term contributes with a positive sign in (6). Using (1), (5), and (6) one obtainṡ…”

“…Here the main challenge consists in the proof of stability required to guarantee the boundedness of all signals and states during the continuous parameter adaptation. An auto-tuning feed-forward friction compensation using the gradient-based algorithm and LuGre friction model has been proposed and experimentally evaluated in [6]. However, the special adaptation gains and predefined reference signals have been presupposed to restrict the parameters within a domain.…”

A novel robust disturbance compensation scheme for d.c. servomotors

“…If all the inertia is reflected to the motor shaft and the armature inductance L is neglected, the transfer function can thus be expressed as equation (4) under the assumption of negligible frictional torque. To estimate the transfer function in this study, a band-limited signal biased as the excitation command u, as defined in reference [30], was specified, i.e. Figure 5 shows the excitation command u and the output response.…”

Online friction parameter estimation for machine tools

“…Linearization‐based methods are easy to implement but the validity of the results can be limited to a narrow region around the operation point. A method for online identification of the LuGre friction model was demonstrated in Reference . A restriction of this approach is the requirement for a closed‐loop scheme with feed‐forward friction compensation.…”