Large synchronous generators play an important role in the power generation industry and its duty in this way is indispensable. Although synchronous generators are reliable, they are subjected to some modes of failures. If a generator continues to operate under faulty conditions, its efficiency drops considerably and its life span is shortened. As a matter of fact, the generator stoppage and outage from the energy generation process causes huge economic loss. Consequently, fault diagnosis at initial stages of occurrence not only prevents the fault extension, high periodical expenses and outage of the generation loop but also preserves the nominal life of the generator. These failures in synchronous generators may be inherent itself or due to the operating conditions.

Faults in synchronous machines include: static, dynamic, and mixed eccentricity, stator, and rotor Interturn and ground short circuit fault, broken damper, and end ring fault. Few methods have been proposed and applied in order to detect several kinds of failures in synchronous generators at different stages, however, most of them are unsuccessful in detection procedures. The basis of any reliable fault diagnosis method is the inclusion of the real behavior and conditions of the faulty machine. Consequently, a proper and at the same time the most important step will be modeling of the problem. Modeling method is the foundation for the next step of fault detection. In this study, Ansys Electronic will be used as a way to simulate a synchronous machine in a healthy and different type of faults from no-load to full load. Experimental results should be used to certify the simulation results. For this purpose, experimental set up which is provided with different types of fault like Static Eccentricity, Excitation short circuit fault, and broken damper bar fault.

The main goal of this project is based on the fact that new indices should be introduced in order to detect the fault at its early stage, as a matter of fact, novel theoretical indices for eccentricity short circuit and broken damper faults based on magnetic flux, electromotive force and vibration should be extracted. These analytical indexes should be precise enough for fault detection purpose therefore, the saturation and stator slot effects should be taken into account.

The output of the FEM simulation of the synchronous generators such as current, electromotive force, magnetomotive force, air-gap magnetic flux, vibration and shaft fl ux, and voltage should be analyzed using
time or frequency domain based processors. Novel theoretical indices should be demonstrated in the processed signals of nominated processors. Finally, classifier and artificial intelligence tools should be used to discriminate the severity and type of faults.