الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 164 |  |  |
| | | from | 164 | | |
Abstract
Permanent magnet synchronous motors (PMSMs) are widely used in many industrial applications for their high efficiency, compact structure, simple mechanical construction and fast dynamic response. As their cost continues to decrease, they have the opportunity to become a dominant force in the industrial applications market. One of the most important problems in permanent magnet motors is the torque ripples, which is inherent in their design. These ripples are parasitic, and can lead to mechanical vibrations, acoustic noise, and problems in drive systems. Minimizing these ripples is of great importance in the design of PMSMs.
There are three sources of torque ripples coming from the machine: a) cogging torque, b) difference between presences of the air gap in the d- and q-axis (reluctance torque), and c) distortion of the magnetic flux density waveform in the air gap. Minimizing torque ripples is of importance in many industrial applications and this is the reason which has received much attention in recent years. Different techniques for torque ripples minimization have been proposed in literature
In this thesis four approaches are proposed for torque ripples minimization in PMSM such as: proportional - integral control (PI), adaptive control and two methods using artificial neural networks (ANN). The proposed PI method is based on PI current controllers enabling tracking of quadrature current command values, the proposed adaptive method is based on two loop controllers (current controller and speed controller) in addition to using space vector pulse width modulation to maximize fundamental component of torque. The first method of ANN is based on two loop controllers (current controller and speed controller). The second method of ANN is based on estimation of torque constant and stator resistance in PMSM. The q-axis inductance is modeled off-line according to q-axis stator current. The neural weights are initially chosen small randomly and a model reference control algorithm adjusts those weights to give the optimal values. The ANN parameter estimator has been applied to flux linkage torque ripple minimization of the PMSM.
Comparative analysis proves the effectiveness of the suggested adaptive controller and two methods of ANN than the classical PI one according to ripple reduction as well as dynamic response. Moreover, the comparative results of the Adaptive Control and two methods of ANN illustrate the reduction in the torque ripples compared with previous related work. The validity of the proposed methods is confirmed by simulation results. The simulation results show an improvement using neural network based estimator for torque constant and stator winding resistance, over the other controllers in torque responses. The four approaches are explained in clear details, which are designed using SIMULINK under Matlab /2009 software package.