Dr. Abduladhem A. Ali, Dr. Abaas H. Abaas & Ahmed Thamer Radhi International Journal of Artificial Intelligence and Expert Systems (IJAE), Volume (3) : Issue (1) : 2012 14 Differential Protection of Generator by Using Neural Network, Fuzzy Neural and Fuzzy Neural Petri Net Prof. Dr. Abduladhem A. Ali abduladem1@yahoo.com College of Eng. /Computer Dept. University of Basrah Basrah, Iraq Asst. Prof. Dr. Abaas H. Abaas abbashafeth@yahoo.com College of Eng. /Electrical Dept. University of Basrah Basrah, Iraq Ahmed Thamer Radhi dr.ahmed197750@yahoo.com Technical Institute/Al-Amarah/Elect. Dept. Foundation of Technical Education Missan, Iraq Abstract This paper deals with the applications of Artificial Intelligence techniques for detecting internal faults in Power generators. Three techniques are used which are Neural Net (NN), Fuzzy Neural Net (FNN) and Fuzzy Neural Petri Net (FNPN) to implement differential protection of generator. MATLAB toolbox has been used for simulations and generation of faults data for training the programs for different faults cases and to implement the relays. Results of different fault cases are presented and these results are compared among the three implemented techniques of relays and with the conventional differential relay of generator. Keywords: Differential Protection, Generator Internal Faults, Neural Net, Fuzzy Neural and Fuzzy Neural Petri Net. 1. INTRODUCTION Synchronous generator is the most important element of power system. Generators do experience short circuits and abnormal electrical conditions. In many cases, equipment damage due to these events can be reduced or prevented by proper generator protection. Generators need to protect from abnormal conditions, when subjected to these conditions, damage or complete failure can occur within seconds, thus requiring automatic detection and tripping. All faults associated with synchronous generators may be classified as either insulation failures or abnormal running conditions [1, 2]. An insulation failure in the stator winding will result in an inter-turn fault, a phase fault or a ground fault, etc. At present the generators are protected against almost all kind of faults using differential methods of protection. Differential relays, in particular the digital ones, are used to detect stator faults of generators. Electric power utilities and industrial plants use electromechanical and solid-state relays for protecting synchronous generators [3]. With the advent of digital technology have made significant progress in developing protection systems based on digital techniques [4,5]. Protection relaying is just as much a candidate for application of pattern recognition. The majority of power system protection techniques are involved in defining the system state through identifying the pattern of current waveforms measured at the relay location. This means that the development of adaptive protection can be essentially treated as a problem of pattern recognition. Artificial Intelligences (AIs) are powerful in pattern recognition and classification. They possess excellent features such as generalization capability, noise immunity, robustness and fault tolerance. AI-based techniques have been used in power system protection and encouraging results are obtained [6, 7]. Artificial neural network is a kind of network structure based on modern biology nervous system research, which shows great application potential on equipment diagnosis by its capabilities of parallel distributed processing, associative memory and self learning. Through learning on multiple types of fault samples, a single NN can memorize characteristics of such faults, thus a single NN can