Using feed-forward back propagation for Rotor Flux Estimation of a Bearingless Induction Motor applied in the Speed Vector Control

This study presents the problem of rotor flux orientation control of induction-type bearingless motor. The key of this solution is the estimation of rotor flux. The neural network is able to estimate accurately the rotor flux magnitude or position. The bearingless induction motor model is used to obtain the training data and the learning technique used was investigated by computer simulation. The bearingless induction motor model characteristics were 3,75 kW, two pole-pair, 60 Hz, air-gap length 0.2 mm cage rotor is based on an input-output model. The adopted model have balanced three-phase currents and despised the viscous friction of the bearings. The software environment used for this simulation was MATLAB® R2010a. The motor equation were solved by using step- by-step numerical integration with an integration 10 -5 s. The simulated results showed good performance. It was used a simulator based on the finite elements method for acquiring flux density for Bearingless Induction Motor model. This paper aims at compensating possible parametric variations of the motor caused by agents such as temperature or nucleus saturation and that neural network flux estimation may be a feasible alternative to other flux estimation methods. The results obtained by simulation confirm the effectiveness of the method.