Gain Property and Data Analysis for Diagnosing Failures in a High-Efficiency Induction Motor

Induction motors are the most commonly used in the industrial market, corresponding to 90% in areas such as manufacturing, pharmaceutical, machines and tools; this is due to its robustness compared to other types of machines. Due to the main role they play in large scale production, they should not stop due to failures. From this perspective, it is intended to diagnose any type of malfunction that occurs in these traction machines, before a production stop takes place. These situations give rise to the proposition of a variety of time-domain and frequency-domain methods to make a successful diagnosis of the failures. This paper proposes the Gain Property method, which relates the currents and voltages (C/V) supplied to a high-efficiency induction motor; the results obtained by such method are stated in two ways: using statistical tools (gray correlation, average deviation and quadratic deviation) and a Bayesian probabilistic tool, in order to analyze the behavior of the results and obtain a favorable diagnosis. In a testbench the motor was subject to four types of incipient failures, and after processing the data of the gains in the three supply lines it was concluded that, depending on the techniques used as statistical tool, the effectiveness of the diagnosis changes, approximating its results in 35%; on the other hand, the Bayesian probabilistic method exhibited a significant improvement for failure diagnosis.