It really comes down to reducing current in the motor, faster acceleration will get the motor moving and current will drop. I’m sure it’s possible to model the motor and estimate current and heating, but as you can receive both as status from the motor it may just be easier to do some experiments.
This reminds me a bit of the work Chris and I did on modeling PTCs and the smart motor library all those years ago.
( most links will be broken at this stage, too many forum migrations