In 1979, D.C. Lee et al., 5 propose a LOE detecting way which have been simulated on nuclear power plants which can sense the DC field voltage and stator terminal voltage variation during a failure in excitation system. The method was tested only on short circuit fault in the excitation system without considering the other types of excitation system faults, it does not sense open circuit fault in the excitation system 1. Also, it fails to differentiate excitation loss and system failure like stable power swing which happen due to faults in the transmission line.
In 1949, Mason 6 suggests a negative off-set mho-type distance relay to sense the impedance variation of generator terminal point due to excitation loss. When the impedance falls under predefined protective zones in R-X plane for a pre-set time delay determined using the longest oscillation of swing angle, the relay detects loss of excitation and send a trip signal to the field breaker. Even if this is the basis for most of the methods that have been created since, but it has high relay operation time and difficulties in differentiating system failures like stable power swing which are failures outside the excitation system.
Shortly afterwards, in 1975, Berdy 7 presented a method based on the addition of another mho unit to this protection scheme proposed by Mason to decrease the operation time. However, still this method is also based on apparent impedance may mal operate during stable power swings (SPS), such as a severe fault occurrence near to the generator operating with a leading power factor and in partial LOE. In 2009, A. P.
Morais, et al., replace the negative offset unit on the second zone with a positive offset mho unit to improve the sensitivity of the distance relay on partial loss of excitation too 8. Even if still the methods are not practical due to the vast simulation processes, resistance and reactance scheme method, an admittance scheme which uses conductance-susceptance plan, the active and reactive power scheme and other schemes using the generator parameters variation with excitation loss even have been recommended with help of digital relays in 9, 10.
As have been compared in 9, all the above LOE detecting schemes have been found mal operated under system disturbances especially for stable power swings that happen due to a sever faults near to the generator terminal. This leads to unwanted tripping of the field breaker for system disturbances too which will cost even blackout of the whole system 11. In 2005, S. R. Tambay and Y. G. Paithankar 12 proposed to use a relay with quadrangular characteristic and the use of rate of change of the reactance seen in the terminals of the machine with the help of digital relays to improve the accuracy and the performances of impedance type protection 12.
With the same objective A. P. de Morais et al., propose a fuzzy interference method of LOE detection in 2010.
Again with the advent of digital relays, some improvements have been mentioned using different characteristics of the generator on excitation loss case to differentiate the LOE and system disturbances mainly stable power swing. In 14, stable power swing sensing depend on the variation of angular velocity and generator terminal voltage acceleration have been suggested so that SPS is easily discriminate without the dependence of other conditions. An artificial intelligent dependent neural network called artificial neural network (ANN) have been suggested in 15, another method on the basis of Space Vector Machine (SVM) technique to discriminate between LOF and SPS is presented in 16 and decision tree technique in 17 have been proposed on different years. These methods may present good results in face of discriminating LOE and SPS, however they require a considerable amount of training and depend on the characteristics of the system.A combined index based on generator terminal voltage, reactive power and power angle variations is presented in 11, the various simulation trends to analyse the indexes and the dependence of electrical parameters on the system condition makes it difficult to implement it practically and differentiate loss of excitation and stable power swing in some special cases that it fails to sense LOE event. In 2016, Behnam Mahamedi and Jian Guo Zhu, 18 present a setting free approach that the rate of resistance variations with excitation loss is introduced as an LOF detector, which remain negative after LOF occurrence for about 1.
76 seconds which specified using the power swing duration of the generator when a fault is happen near the generator terminal. However, the oscillation nature of the resistance with speed variation makes the method un-functional under high slip frequency systems. In 2014, M. Elsamahy, et al., propose the effect of STATCOM on the LOE relay characteristics through the classification techniques of SVM and an adaptive neuro-fussy methods. The terminal voltage of the generator depends on the characteristics of the transmission line too since it is the voltage delivered to the transmission line.
Through the transducer of the excitation system, the terminal voltage of the generator is return back to the excitation terminal to check the stability of the transmission line. Thus the protective devices to keep the voltage on stability of the transmission line have effect on the characteristics of the LOE relays. The effect of mid-point STATCOM have been studied using FACTS devices. In this paper, a new LOE detection algorithm will be presented depend on the calculation of the rate decay of the generator internal voltage when excitation of the generator lose which is a parameter highly sensitive to LOE event; to increase the accuracy and to improve the performance of the LOE detecting method on the generating units to keep the whole system in stable and reliable by decreasing the mal operation of the relays on the generating units.