Toolbox for Online control and Design of tool wear mechanisms when cutting difficult-to-machine materials (ONCODE)

Purpose and goal:

Development of AI solutions to control and influence the evolution of tool wear (to shape it), predict tool damage and estimate process efficiency. Expected results and effects:

The project addresses the development of a toolbox for an AI-based platform/demonstrator of PCM when machining difficult-to-cut materials (relatively expensive materials for responsible parts, where precision and quality are of vital importance) with applications in aerospace and automotive industries (Ti- and Ni-based). The developed solution(s) will also be of great interest to tool manufacturers in the form of a recommender system for customers with different needs.

Approach and implementation:

Using Reinforcement Learning (RL) terminology, the problem statement can be formulated as follows: development of the agent which consists of the interacting AI-based Digital Twin (DT) of the process, TCM, and Decision Making (DM) blocks reacting on the lubricant/coolant supply and estimating process efficiency through the observations obtained by the array of sensors. Non-RL solution will look like several interacted AI solutions (TCM - DT - DM) integrated into the PCM system.