Reinforcement Learning For Dynamic Machining Process Parameters

This project is supported by the Department of Energy

In this work we use offline reinforcement learning to find optimal dynamic process parameters (speeds and feeds) while machining alloys.

This work pays attention to the effects of tool-workpiece interaction on workpiece quality and overall energetic costs.

Outcomes:

1. Applied this process to TiAl and Inconel to improve productivity by more than $100$%, leading to improved profitability and lower embodied-energy.
2. Reduced the entire pipeline consisting of characterizing material to testing a dynamic processing strategy from months to weeks