Project Details
Projekt
Real-time capable methods for control
Applicant
Professor Dr.-Ing. Martin Mönnigmann
Subject Area
Automation, Mechatronics, Control Systems, Intelligent Technical Systems, Robotics
Term
since 2025
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 512940050
The aim of the subproject is the development and implementation of fundamental methods for the control of the welding bead generation in laser metal deposition. The two main challenges are (i) ensuring the geometry of the melt pool that determines the welding bead geometry and (ii) simultaneously regulating basic material properties of the printed structures. The desired material properties are to be achieved by controlling the cooling behavior of the deposited material. Initially, an existing nonlinear dynamic multivariable model will be extended and tailored to reliably describe the required state variables and, at the same time, be suitable for real-time use with sampling times in the range of 5-10ms. Based on this model, a state observer in the form of a nonlinear Kalman filter and a predictive controller based on real-time optimizations will be developed. The nonlinear Kalman filter will fuse sensor data from the multimodal sensors, which are already available in dedicated large-scale lab in the federal Research Building ZESS at RUB. The predictive controller is chosen for its suitability for the given multi-input-multi-output system and its nonlinear couplings. Because it is based on real-time optimization, the predictive controller will naturally be able to operate with lower and upper bounds on temperatures and cooling rates. It is essential to permit these process parameters to vary within bounds for increased flexibility and to avoid under-actuation. Recent experimental results indicate melt pool size controllers for laser metal deposition on flat surfaces also work in free-form manufacturing (even up to printing "overhead", i.e. in opposite direction to gravity). The process parameters, specifically the material supply rate, the printing head velocity and the laser power, must be confined to bounds that depend on the angles of the material deposition with respect to gravity and the existing material for this purpose. By virtue of the optimization-based controllers, these constraints can seamlessly be integrated even if they need to be time-variant. An embedded system based on an industrial PC will be set up for real-time implementation of the observer and controller algorithms. Achieving real-time capability of the system and algorithms will require a combination of advanced techniques, such as semi-analytical solutions to heat conduction equations, move blocking and constraint removal.
DFG Programme
Research Units
