Estudo comparativo de técnicas para Diagnóstico de falhas em motores de Indução trifásicos

Abstract

Electrical motors are responsible for 95% of primary movement source in industrialized nations. Among those, 90% are Three-Phase Induction Motors, present in almost all industrial sectors. Due to its importance on this sector, there is a need for monitoring them in order to avoid production stops and operational disasters. In this work, studies were conducted on common fault diagnostics in Three-Phase Induction Motors intending industrial applications. Different sensoring techniques were used and their performance were analyzed. An embedded system was developed to make field applications with different techniques possible. This installation can be done noninvasively and data collection can be obtained in real time. Fast Fourier Transform (FFT) and wavelet processing techniques are used as tools in mathematical processing of the data. In the first moment, fault analyses were conducted offline making use of data acquisition devices and further processing of the information. In a second phase, the embedded system was used to monitor automatically (online) the evolution of the damage developed. The system receives the motor current signal, and using local processing, conducts the spectral analysis of the signal identifying incipient faults. These data are available for communication with or without wires. For the embedded system implementation, the algorithms were adjusted to comply with the embedded hardware resources restrictions. Through theoretical and experimental development, several techniques were used and compared with the objective of performing a full diagnostic TIM malfunction. The experimental results corroborate the theoretical ones, and it was conducted a detailed study of the methods on the state of the art and new approaches were made.