Model based predictive force control for multi-axis milling (MPFC)

Key Info

Basic Information

01.05.2021 to 30.04.2024
Organizational Unit:
Chair of Manufacturing Technology, Cutting Technology
German Research Foundation DFG

Research partner

    • Institut of Automatic Control (IRT), RWTH Aachen University



Adrian Rüppel



+49 241 80 28020



Industrial machine tools control machine variables (position, speed, acceleration) in cascaded control loops. The process itself remains outside and is only rudimentarily monitored. The machine as a controlled system can be modeled and reliably controlled by conventional, reactive control methods (PID controllers).

The control of process instead of machine variables offers several advantages. In numerous research studies, the process force has been shown to be a variable that has a great impact on the safety and productivity of the process, as well as being a suitable measurement variable. The advantages of a force control in milling are:

  • Optimizing productivity in terms of process time and material removal rate
  • Enhancing tool live by ensuring a constant load during different engagement simulations
  • Considering material differences and tool wear

Controlling a milling process instead of a machine tool also comes with new challenges. Milling as a very flexible process is characterized by constantly changing engagement conditions between the tool and the work piece. This results in a constantly changing system to control and, therefore, proofing classic reactive controllers as not applicable.

Advanced control systems shifting the system model from the planning phase to the online phase, resulting in a constantly changing model, which is identified online. At the WZL of RWTH Aachen University, in a cooperation with th Institute of Automatic Control (IRT), the model predictive control (MPC) is researched to control process forces in milling. A MPC is constantly solving a restrained optimization problem to predict the short-timed engagement situation. By using a machine and process model, it is able to optimize the feed of the process, so that the process forces can be kept at a constant level. To close the control loop, the process model is constantly identified online, so that tool wear and material inhomogeneities can be considered within the system.

In past projects, an MPC has only been researched in three-axis milling processes. This project investigates the extension of the system to the multi-axis milling process.