الفهرس 
Power ConvertersOnly 14 pages are availabe for public view
 |  | from | 153 |  |  |
| | | from | 153 | | |
Abstract
The modern high quality permanent magnet materials have enabled us to achieve compact motor designs. Accelerated achievements in the fields of microprocessor-based controllers and DSPs, modern achievements in the fields of power electronics - in both devices and converters - and the evolution of modern analysis of electrical machines have revolutionized the permanent magnet motors AC drives. Permanent magnet motors AC drives have increasingly become competitive in both servo drives and industrial applications as well. These motors are mainly applied to replace the conventional brushed permanent magnet DC motors. The main interest of this thesis is to develop a complete Interior Permanent Magnet Synchronous Motor drive system. Complete mathematical model of the machine is presented. The specific reasons behind selecting rotor reference frame to refer the machine nonlinear differential equations are demonstrated. The application of maximum torque per unit current criteria is presented. Maximum torque per unit current criterion is derived in details. The advantages of the application of this criterion are verified.
The essential tasks of the current controller are investigated. Fixed band hysteresis, sinusoidal band hysteresis and Space Vector Modulator current controllers are presented. The effect of both hysteresis band and the switching frequency on the quality of the current wave is demonstrated. The effect of the switching frequency on the inverter modes of operation is investigated.
Theoretical background of the artificial neural networks (ANN) is presented. Theoretical concepts about nonlinear system identification and the different models of nonlinear systems are also presented. The application of ANN to identify of IPMSM drive system is demonstrated. Application of ANN as a new tool in the field of speed control to compensate the well-known demerits of conventional PI speed controller is being investigated in this thesis. Multilayer feed forward neural network; employing error back propagation algorithm is being applied. The input vector to the neural network speed controller is comprised of a time history of the motor angular velocity. The on-line training algorithm is applied on the output domain. The effect of adding proportional gain Kp in parallel with pure ANN-based speed controller is investigated. The drive system, employing ANN-based speed controller is subjected to all kinds of dynamic performance tests. The robustness of ANN based speed controller against machine parameters variations is explored. Simulation results are presented to indicate the effectiveness of the proposed speed controller.