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

A new incthod Tor tlic detection of the type ui a f:tult in generator circuits and transmission systcms is introduced. Already within a quarter of a cycle aftcr fault inception the method can distinguish between the varions fault types. Pitting the parameters of a set of simple cquations to voltage and currcnt measurements immediatcly before and after a fault identilies the fault type. The procedure includes a new mcthod for phasor computation and takes less than 1 ms coinputation time. As a variant of this mcthod neural nctworks arc employed. Verification using EMTP modeling proved satisfactory opcration of both mcthods even when the current signals were superimposcd with heavy noise. Fast decisions for singlc polc tripping and a crucial basis for algorithms lor synchronous switcliing under fault conditions arc provided.