Model predictive control structures for high-precision rack-and-pinion drives

• Applicant: Professor Dr.-Ing. Alexander Verl
• Subject Area: Production Automation and Assembly Technology
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 556052806

The manufacturing quality and dynamics of production systems are largely determined by the feed drive system used. For machines with long travel distances, rack-and-pinion drives (RPD) are preferred. In order to compensate for the negative effects of the backlash between rack and pinion, electrically preloaded drives are often installed. Due to the constantly increasing demands on the achievable path accuracy and feed dynamics, the previously used cascaded control loop structure consisting of current, velocity and position control is increasingly reaching its dynamic limits. Alternative control methods provide a possibility to improve the tracking and disturbance behavior. The research project will investigate the potential of model predictive control for electrically preloaded RPDs compared to the cascaded controller established in the industry. For this purpose, different model predictive control structures will be used and the respective control performance will be evaluated. Three structures will be considered for the control of an electrically preloaded RPD. In one structure, the P-position controller is substituted by an MPC. Another structure replaces the subordinate PI-velocity controller by an MPC and keeps the P-position controller. A third structure consists of substituting both control loops (P-position and PI-velocity) by an MPC. Previous work has shown the potential of model predictive control for ball screw drives, but the extent to which the RPD specific characteristics of the controlled system (backlash, transmission error, preload) affect the potential in terms of improving the dynamic behavoir (reference tracking and disturbance behavior) of MPCs has not yet been investigated. The considered structures are first fundamentally analyzed in a simulation model considering the prediction models, the cost function and the controller setting. Afterwards, the developed control approach, which is novel for ZRA control, and the methodology will be implemented and experimentally validated on a test bench.

• DFG Programme: Research Grants
• Co-Investigator: Dr.-Ing. Armin Lechler