PMSM control based on fuzzy logic controller modelled with artificial neural network

The Permanent Magnet Synchronous Motor (PMSM) has been produced at lower costs with the innovations in magnet technology and has become widely used in many industrial applications due to its low energy consumption. The widespread use of PMSM has brought along the need for control performance with high level of accuracy. The fact that modern control methods exhibit a more effective control performance than traditional control methods has made the use of modern control methods widespread and studies on modern control methods have intensified. Fuzzy Logic Controller (FLC), which is one of the modern control methods, is an effective control method that allows the system designer to reflect the similarity in human thought structure and the expertise of the system designer to the control system more effectively. Although FLC offers an effective control performance, its response time can be slow due to the complexity of its structure and the intensive mathematical operations it contains. Artificial Neural Networks (ANN) is a very suitable method for system modelling and has a very fast response time thanks to its parallel processing capability. In this thesis, it is aimed to create an ANN model with a faster response time by modelling the FLC, which has an effective control performance, with ANN and to apply the obtained model on the Field Oriented Control (FOC) of PMSM. A data set to be used in the training of the ANN was created with the data taken from the inputs and outputs of the FLC. The ANN model trained with the generated data sets modelled the BMD with a very high accuracy and produced the same output as the FLC for the same inputs. The units required for the PMSM design and the FOC control were designed in the Simulink plug-in of the MATLAB programme. The FLC was designed in the Fuzzy Logic Toolbox plugin of MATLAB and the ANN was designed in the Neural Network Toolbox plugin of MATLAB and transferred to Simulink environment and their performances were compared in simulation studies. In the simulation results, it was observed that FLC exhibited a more successful control performance than the traditional control method, Proportional-Integral (PI) control. It is also observed that the FLC model created with ANN has a faster response time than FLC. This proposed method can be used for PMSM control where a more effective and high speed control performance is required.