Comparison of BPN, RBFN and wavelet neural network in induction motor modelling for speed estimation

Abstract

This paper discusses the question of estimating a neural network induction motor speed. The neural network is trained in the relationship between stator currents and rotor speed. Rotor currents and speed are created using MATLAB from simulated model induction motor and from experimental real-time set-up using LABVIEW. With these two data sets, Induction motor is modelled using three neural network architectures, namely BPN, RBFN and Wavelet Neural Network, comparing the performance of each neural model in estimating speed. Results show the neural approach's ability to replace the speed sensor used in the closed loop speed control system to accurately estimate speed.