Validation
A Federated Learning System Platform Development for CNCs
The AI4CNC experiment aims at building a CNC manufacturing data space for TEKNOPAR’s one of the CNC facilities in İzmir, where 3 CNCs are used to produce metal hydraulic blocks and cylinders. The primary objective of the experiment is to enable federated learning AI models in order to estimate CNC’s tool wear, so that the operator’s decisions are supported by AI about when to replace and what to do regarding the used tools. The development of federated learning will enable data sharing space that also compliance for secure data ownership. AI4CNC will enable digital platform building and data space sharing at TEKNOPAR’s CNC site and enhance the knowledge transfer regarding the of such spaces, offer AI-driven application, to estimate tool-wear of the CNC cutting edges, as an example that is based on the platform and data. The cross-silo federated learning can later be disseminated to cross-device federated learning.
The developed solution may later be solved to other CNC owners. Because the tool-wear is important for CNC product quality and cost related variables. The solution is aimed at being marketed in digital platforms, where support of AI REGIO may also be very beneficial.