A learning method for use in intelligent computer relays for high impedance faults

An attempt is made to find an intelligent detection system. This intelligent system has two specific properties. First, it is capable of taking advantage of multiple detection parameters or variables. Second, it provides adaptability using inductive reasoning, incorporated with some critic modules in a detection system. Inductive reasoning which will minimize entropy was used to acquire the knowledge of fault class and nonfault class. This knowledge was used to make the intelligent system adaptive to surrounding environments. A learning detection system was implemented with inductive reasoning and an event detector. An example execution is shown with a decision rule which was derived using training sample data consisting of high impedance faults, switching events, and normal status. A complicated set of test data is used to test the performance of the learning detection system. It is found that, even when it met very complicated situations, the learning detection system made smart decisions and evolved to a new situation with a newly derived decision rule