Neural network-based stator voltage compensator for low-frequency operation of a vector-controlled induction motor drive

The stator field-oriented sensorless vector control method is receiving wide attention from the standpoints of being sensorless and less dependent on the motor parameters. However, this method has a problem at low speeds because actual stator voltage does not coincide with the corresponding reference voltage owing to the quantization error in the controller, the discrete-time process, the forward voltage drop of switching devices, and the dead time of the inverter. In this study, an artificial neural network-based voltage compensator is proposed to reduce the voltage error between the reference and the actual voltages. The proposed method was performed on an experimental system and the results were found to be excellent.