Statistical Analysis of Multidimensional Components for the Diagnosis of Faults in Electric Motors

Abstract

The current manufacturing technification has resulted in business competitiveness because their products might be customized in some cases, the reason to understand that the production line should not stop in the presence of possible mechanical or electronic failures. Against this background and knowing that approximately 80% of the induction motors operate in the industrial sector, a maintenance record or control should be kept due to its direct relationship with production. With this perspective, various studies attempt to diagnose a type of incipient failure that may occur, traditional and/or robust. Thus, this document employs the multidimensional technique or Park demodulation, to perform an analysis in three case studies, the Park Vector Module (PVM) and each constituted component (1 <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">a</inf> and I <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">d</inf> ). These failures will be identified using statistical tools, analyzing their behavior and indications that reveal a diagnosis in the presence of possible incipient failures; in addition, these results will be helpful for those studies that involve probability or artificial intelligence topics for preventive or predictive diagnoses. Results show that there is evidence for the identification of different types of incipient failures, depending on the type of tool and the analysis case; specifically, one type of failure is identified for the standard deviation, three for box plots and ambiguity generation for the correlation coefficient; however, four types of failures proposed for this case study are identified for the quadratic mean.