Sensorless position estimation of permanent magnet synchronous motors using artificial neural networks

Recently DC and AC motors have replaced by permanent magnet synchronous motors (PMSM) due to their simple and lightweight structure, small dimension and high efficiency. Therefore, they have been widely used in industrial applications especially as servo motor.Rotor position and speed are required to control PMSMs. The encoder used for this purpose has some disadvantages such as higher system cost, larger motor dimensions and decreased safety. Therefore, eliminating these sensors is an important research subject.In this thesis, the sensor and sensorless control of PMSMs was examined in detail. For this aim, the field oriented control with hysteresis and pulse width modulation of PMSM was carried out. In the field oriented sensorless control, the position and speed of rotor were estimated by the artificial neural network estimators the structure of which were formed by a model-base estimation method. MATLAB/Simulink simulation model and experimental application setup were prepared for the sensor and sensorless field oriented control of PMSM. Application circuit was run in real time by using DSPACE 1103 control. The accuracy of the system was proved by comparing the results from simulation model and application (experimental) circuit. It was seen that the position estimations by artificial intelligence observer had a high accuracy especially in wide speed range.