Inertia and load torque estimation in speed controlled electric drives

Abstract

This work deals with the estimation of the inertia and the load torque. Adaptation of the speed control based on the estimation is also discussed.

If the mechanical system of an electric drive is known, the speed controller can be correctly tuned and a good speed response achieved. Often the value of all mechanical parameters, for instance the inertia, is not known. Furthermore, the inertia might vary during operation making the speed control more difficult. Load variations cause speed errors that the speed controller itself does not always remove as fast as desired.

The recursive least-squares algorithm (RLSA) and two slightly different neural networks methods have been used in the estimation. All methods give a good estimate of the inertia when no load is applied. Having a nonzero load, the RLSA is the only method that works properly. The neural networks methods have numerical problems and do not notice changes in the inertia as the load differs from zero.

Simulations where the speed controller was adapted to the changes in the inertia show that the speed response is better. If the speed controller is adapted fast enough, overshoots can be removed from the speed. The response to load torque changes has been improved by load torque feed forward.