Metodologia para predição de tempo de falha de máquinas e equipamentos baseada no monitoramento de vibração

Abstract

The purpose of this work is to propose a methodology for predicting machine and equipment failure time based on vibration monitoring. Several measurements have been made in an experimental workbench, which was composed of a motor and five bearings. The objective of these measurements was to analyze the vibration data collected by applying specific signal analysis tools. Those specific tools consist on calculating vibroacoustic symptoms, whose temporal evolution has been evaluated. The objective was to verify which symptom (s) could be modeled according to the Weibull distribution, which is widely used to evaluate equipment lifetime. Among the symptoms evaluated, the following symptoms could be modeled according to the Weibull distribution: kurtosis, skewness, K4, TDA with high pass filter of 3000 Hz, energy level of the envelope with low pass filter in 100 Hz, energy level of the filtered envelope with bandpass filter between 1000 and 2000 Hz, filtered envelope energy level with bandpass filter between 2000 and 4000 Hz.