Teste de hipóteses sobre o espectro de freqüência, aplicado na manutenção preditiva de motores de indução

AUTOR(ES)
FONTE

IBICT - Instituto Brasileiro de Informação em Ciência e Tecnologia

DATA DE PUBLICAÇÃO

21/08/2008

RESUMO

This project had begun with the aim of technical improvement and a predictive maintenance tool development for electrical machines diagnosis. Based on nonintrusive approach Motor Current Signature Analysis - MCSA for three-phase induction motor broken bars detection, it was developed a spectral analysis method able to increase accuracy and reliability, compared to simple Fast Fourier Transform - FFT use. Upon FFT stochastic properties and deficiencies analysis, here presented, it was developed a Maximum Likelihood Estimator - MLE that allows Hypothesis Test implementation for Gaussian random variable from signal s amplitude, phase and continuous frequency estimates. To checkup its viability, this method was implemented for broken bars detection and achieved results as efficient as traditional analysis does. In addition, the method returns a suitable diagnostic variable for fault progress analysis, customized for each motor specifics characteristics by means of confidence intervals determination for diagnostic. Analysis under commonly industrial environment disturbance (under voltages, over voltages, unbalanced phases) and other sort of mechanical and electrical faults (axis unbalance and misalignment, single phase interturns short-circuit) tests revealed conditions in which traditional analysis variable (usually presents in artificial intelligence methods) may be changed leading to jeopardize the diagnostic performance.

ASSUNTO(S)

engenharia elétrica teses.

ACESSO AO ARTIGO

Documentos Relacionados