Two new methods for very fast fault type detection by means of parameter fitting and artificial neural networks

Summary form only given as follows. A new method for the detection of the type of a fault in generator circuits and transmission systems is introduced. Already within a quarter of a cycle after fault inception the method can distinguish between the various fault types. Fitting the parameters of a set of simple equations to voltage and current measurements immediately before and after a fault identifies the fault type. The procedure includes a new method for phasor computation and takes less than 1 ms computation time. As a variant of this method neural networks are employed. Verification using EMTP modeling proved satisfactory operation of both methods even when the current signals were superimposed with heavy noise. Fast decisions for single pole tripping and a crucial basis for algorithms for synchronous switching under fault conditions are provided.