A SELF-INSTRUCTING EXPERT-SYSTEM FOR CONTROLLING ELECTRICAL-POWER SYSTEMS IN EMERGENCY CONDITIONS

The paper describes a self-instruction procedure for recognition of el ectrical power system (EPS) conditions with large disturbances and giv es algorithms for choosing control actions in expert systems. It is sh own that a self-instructing expert system for recognition of EPS condi tions and for their control leads to high accuracy of pattern recognit ion. It also reduces the volume of condition information to be process ed (in the ''BEFORE'' mode) and is characterized by fast action and a small volume of computer memory. This allows automatic processing of t he information received on the state of the EPS and uses pattern recog nition analysis to correct the previous information, i.e. to improve t he set of EPS instructing conditions.