Using an Artificial Neural Network as a Rotor Resistance Estimator in the Indirect Vector Control of an Induction Motor

This paper presents a rotor resistance estimator based on an artificial neural network (ANN) used in the indirect vector control (IVC) of an induction motor (IM). Attention is focused on the dynamic performance of ANN rotor estimator, which gives superior performance over the fuzzy logic based rotor estimator reported in technical literature. The simulation was done using a 1.5 HP induction motor. The same ANN rotor estimator was proved with other IM having different rated power. The use of the same ANN was possible because the scaling and descaling (normalization) of the input and output of ANN was property done for each motor. The ANN training was done offline using the Levenberg-Marquardt algorithm. The neuronal network is a three-layer network; the first layer has fourteen neurons (or nodes), the hidden layer has five neurons and the output layer has only one neuron because the unique output signal is the rotor resistance value.