On-line speed estimation based on ANN for PMSM sensorless speed control

  • Authors:

  • Bassel Sahhary;Hossam Abbas

  • Affiliations:

  • Helmut- Schmidt-University/University of the Federal Armed Forces, Hamburg, Hamburg, Germany;Hamburg University of Technology, Hamburg, Germany

  • Venue:
  • MIC '08 Proceedings of the 27th IASTED International Conference on Modelling, Identification and Control
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper proposes a method for on-line speed estimation of a permanent magnet synchronous motor (PMSM) based on an artificial neural network (ANN), where, the conventional Model Reference Adaptive system (MRAS), which is usually used to estimate the rotor speed of the PMSM, is represented by an ANN. The ANN contains adjustable and constant weights. The adjustable weights are proportional to the rotor speed. The adjustable weights are changed by using the error between the outputs of the reference model and the ANN, which is the PMSM's output active power, since any mismatch between the actual and the estimated rotor speeds results in an error between the mentioned outputs. The steepest decent method is used to adjust on-line the weights of the ANN. Experimental and simulation results illustrate the practicality of the proposed method.