PREDICTION OF INCIPIENT FAILURE OF ELECTRIC POWER SYSTEM COMPONENTS.

An investigation of a real-time system which could predict certain types of incipient failure of electric power-system components is described. The system is based on the use of simple and inexpensive voltage and current sensors, a microcomputer, and a set of signal processing algorithms. A decision-theoretic method of pattern recognition is employed to recognize and classify incipient failure patterns. After a brief overview of the experimental system, this paper focuses on the mathematical basis for the incipient failure prediction using only terminal voltages and currents.