Institute of Production Engineering and Machine Tools Institute News and Events News
Intelligent manufacturing: How smart is process planning getting?

Intelligent manufacturing: How smart is process planning getting?

© IFW
Learn WZS: Project progress for self-optimizing process adaptation for tool grinding

15 months ago, the LearnWZS project was started, in which intelligent process planning for tool grinding is being developed. "We want to combine the data-bound knowledge from machine, process and simulation in process planning and thus offer the user a method for qualitative and quantitative optimization of the manufacturing process," says Michael Wulf from the Institute of Production Engineering and Machine Tools (IFW) at the Leibniz Universität Hannover, summarizing the project goal. The aim is to reduce run-in processes for the design of new manipulated variable combinations in order to save resources such as the raw material or machine occupancy.

The world of manufacturing technology offers more and more possibilities for data acquisition and processing in connection with advancing digitalization. Especially for the process planning of complex kinematics with process parameters at the edge of process reliability, this data basis is essential to overcome the conflict between quality and quantity. An application example represents tool grinding as a cost and time intensive manufacturing process. The synchronous movements in 5 machine axes, combined with the meshing situation that is difficult to grasp analytically, require a high level of experience and process understanding from the process engineer. In addition, the raw material is very valuable due to the high demands on the cutting materials, which is why scrap is a significant cost factor.

Until now, costly and time-consuming run-in tests were required for new manipulated variable combinations, which can be reduced with the method to be developed. The final developed and researched method is now much closer after half of the project duration. The first two work packages "data acquisition and fusion" and "learning process models" have now been successfully completed. In the first work package, a process-parallel data acquisition and simulation system was set up to synchronize the various data sources. For this purpose, control data such as axis currents and positions are read out and fed into a machine-specific simulation environment of the IFW CutS software. The simulation parameters cycle time and resolution of the discretization were empirically researched for this purpose and evaluated in a comparison of the simulated with analytical and real measured meshing variables. Subsequently, the simulation could be optimized with respect to the minimum required discretization for a minimum possible computation time.

In the second work package, the process force and the workpiece surface were modeled, allowing the processes inherent in the process to be approximated numerically. This required a method to compare the geometry-dependent engagement situations of the different processes. Surface grinding tests enabled a real measurement of the process forces, which in turn could be assigned to the local engagement variables. Subsequently, the knowledge gained was applied analogously to the kinematically more demanding tool grinding process in order to model the process forces there without force measurement data.

The outstanding work packages three and four comprise the process adaptation and finally the exploration of the overall method developed so far. The process adaptation uses the force modeling of the third work package and optimizes the grinding path with respect to the target variables of workpiece quality and quantity. Prioritization can take place in the direction of one of the two target variables. The last work package will investigate the methods developed so far with regard to their validity for other manipulated variable ranges. In addition to the workpiece material, the grinding wheel bond type and the grain size of the grinding wheel will also be varied. The work package is intended to show how the modeling learns independently and adapts to the changed boundary conditions.

According to Wulf, "At the end of the project, there will then be a method for making the tool grinding machine more intelligent and effectively supporting the process planner. This is a great added value, especially in view of the shortage of skilled workers, to enable even inexperienced employees to plan and manufacture complex workpieces independently." This supports efficient order processing and provides all users with additional security in process planning. The final step is the formulation of a general principle that can be used to plan other manufacturing processes more intelligently on the basis of the method developed.

For further information, please contact Michael Wulf, Institute of Production Engineering and Machine Tools, at +49 511 762 18354 or by e-mail at wulf_m@ifw.uni-hannover.de.