Fault detection in measuring systems of power plants

Abstract

This paper describes possibility of forming diagnostic relations based on application of the artifical neural networks (ANNs), intended for the identifying of degradation of measuring instruments used in developed power systems. As an example a steam turbine high-power plant was used. And, simulative calculations were applied to forming diagnostic neural relations. Both degradation of the measuring instruments and simultaneously occurring degradation of the measuring instruments and thermal cycle component devices, were taken into account. Good quality of diagnostic neural relations was stated. They make it possible to distinguish degradation of measuring instruments from degradation of thermal cycle components. The calculated errors of identification of dergraded devices and measuring instruments in the case of simultaneous occurence of three different degradations were on the level of 0.25 %. Performance of the relations was presented by using an example based on industrial practice.