In this project, as explained above, we develop a PMSM control method by using AI or mathematical model. More concretely, AI and Math will be used for dynamic parameter tuning, tolerance for real machine disturbance/noises, etc. Finally, developed technologies will be applied to various size/type of motor, without using traditional ways such as manual parameter tuning. What we are expecting now is to develop a mechanism that can train AI on real machine (test bench) and measurement data (not data derived by simulator) for practical uses. In terms of methodology, we are now thinking that System Identification, Self-commissioning, Neural network and Reinforcement Learning (of course, we can change to other methodologies after understanding concrete issues).