A comparison of induction motor speed estimation using conventional MRAS and an AI-based MRAS parallel system

Abstract

The Model Reference Adaptive System (MRAS) is probably the most widely applied speed sensorless drive control scheme. This chapter compares induction motor speed estimation using conventional MRAS and AI-based MRAS with Stator Resistance Compensation. A conventional mathematical model based MRAS speed estimation scheme can give a relatively precise speed estimation result, but errors will occur during low frequency operation. Furthermore, it is also very sensitive to machine parameter variations. An AI-based MRAS-based system with a Stator Resistance Compensation model can improve the speed estimation accuracy and is relatively robust to parameter variations even at an extremely low frequency. Simulation results using a validated machine model are used to demonstrate the improved behaviour. © 2009 Springer Netherlands.