Open-loop neuro-fuzzy speed estimator applied to vector and scalar induction motor drives

Abstract

Scalar and vector drives have been the cornerstones of control of industrial motors for decades. In both the elimination of mechanical speed sensor consists in a trend of modern drives. This work proposes the development of an adaptive neuro-fuzzy inference system (ANFIS) angular rotor speed estimator applied to vector and scalar drives. A multi-frequency training of ANFIS is proposed, initially for a V/f scheme and after that a vector drive with magnetizing flux oriented control is proposed. In the literature ANFIS has been commonly proposed as a speed controller in substitution of the classical PI controller of the speed control loop. This paper investigates the ANFIS as an open-loop speed estimator instead. The subtractive clustering technique was used as strategy for generating the membership functions for all the incoming signal inputs of ANFIS. This provided a better analysis of the training data set improving the comprehension of the estimator. Additionally the subtractive cluster technique allowed the training with experimental data corrupted by noise improving the estimator robustness. Simulations to evaluate the performance of the estimator considering the V/f and vector drive system were realized using the Matlab/Simulink® software. Finally experimental results are presented to validate the ANFIS open loop estimator. © 2014 Elsevier B.V.