# How can I model a hysteresis curve from data?

I have the task of modeling the current to torque mapping for a given motor. I have an experimental set up where I can retrieve current, torque pairs.

Now my initial approach was to model the relationship with a regression curve, but I realized that the motor certainly shows some kind of hysteresis.

How would I be able to compute a current to torque mapping that also includes hysteresis from the data I obtained

• Have you tried to graph your data on using tool like excel? Commented Dec 23, 2019 at 0:40
• What kind of motor it is and why should it show any hysteresis? If it is only the motor being tested without any mechanical load hysteresis can be neglected if speed(ie torque) is not changed too abruptly. Commented Aug 19, 2020 at 8:51
• Just take a series of points ramping up, then a second series ramping down. Plot them as separate curves. The hysteresis represents the difference between the two.
– Drew
Commented Jan 7, 2023 at 6:56

## 2 Answers

A very general way to model hysteresis is by a preisach model eg https://en.m.wikipedia.org/wiki/Preisach_model_of_hysteresis it can get very computationally expensive though. Can you post some plots of your data and tell what you want to do with it? That might help get a better answer

This is probably either motor backlash or residual magnetisim in the motor poles, or both.

You'll need to measure out how much is actually there.

There are many ways to model this. The basic idea is to have a piston that fills up and doesn't allow anything to pass unless the piston is either fully extended or fully retracted.