Monitoring of induction motor load by neural network techniques

TLDR: In this paper, the effect of different load anomalies on current spectrum, in comparison with other machine problems like rotor asymmetries, are investigated, and a neural network approach can help the torque pattern recognition, improving the interpretation of machine anomalies effects.

Abstract: This paper deals with the electric tracing of the load variation of an induction machine supplied by the mains. A load problem, like a torque dip, affects the machine supply current and consequently it should be possible to use the current pattern to detect features of the torque pattern, using the machine itself as a torque sensor. But current signature depends on many phenomena and misunderstandings are possible. At first the effect of different load anomalies on current spectrum, in comparison with other machine problems like rotor asymmetries, are investigated. Reference is made to low frequency torque disturbances, which cause a quasistationary machine behavior. Simplified relationships, validated by simulation results and by experimental results, are developed to address the current spectrum features. In order to detect on-line anomalies, a current signature extraction is performed by the time-frequency spectrum approach. This method allows the detection of random faults as well. Finally it is shown that a neural network approach can help the torque pattern recognition, improving the interpretation of machine anomalies effects.