Method of Distance Load Recognition based on the Processing of High Sampled Current Oscillograms

Abstract

This paper is devoted to a new method of load recognition that discards direct installation of sensors near consumer loads (distant sensorless technology). This method can be used for state monitoring of critical electrical loads when the creation of a conventional local dispatch system with traditional sensors is impossible. This method's key idea is to employ a unique combination of discrete and continuous wavelet transforms to analyze the high sampled oscillograms of current loads. The continuous transform applies a set of non-standard consumer's mother wavelets, making it possible to single out one out of identical electrical loads in the complex current signal. This algorithm is pre-configured and trained to identify target loads, and it also has conditions of applicability. It could be part of a software in a device for monitoring critical electrical loads in life support systems (water and air pumps in mines, electric motors operating in aggressive conditions, etc.). In addition, it is possible to control individual malfunctions of electric devices. The criterion of deterioration in stability of load recognition is the decline in the quality of the set of unique consumer's mother wavelets.