249 0

System Fault Diagnosis Based on Transient Variation of Characteristics

System Fault Diagnosis Based on Transient Variation of Characteristics
Issue Date
In this study, wave based fault diagnosis methods in dynamic system depending on model based and signal based methods were suggested. Irregular wave signal has a significant correlation to the possible defects in the system. In order to find such a fault, model-based and signal-based fault detect methods were applied based on the wave signal characteristics. First, as the study of model-based fault detect method, measurement methods to measure frequency-dependent electromagnetic properties of thin film nonconductive medium are proposed. THz wave propagations which is reflected and transmitted are simulated as a pulse wave using Maxwell equation. And the boundary conditions are determined by Fresnel equations and continuity equation. Wave speeds of reflected THz pulse are measured through the transfer function method requiring just a single measurements of pulse wave. Using the developed methods, the complex dielectric permittivity with frequency of materials are measured. Continuous variations of the measured loss factors with frequency were obtained. For the measurement of the THz reflection wave, pitch patch mode is performed. And the signals are settled to a second reflection pulse wave except for the disturbance. The dielectric constant and attenuation values were obtained through numerical analysis in the frequency domain. And then BSR (Buzz, squeak, and rattle) noises classified depending on their temporal and spectral characteristics. Their influence on the acoustic comfort should also be quantified for evaluation. This information is crucial for suggesting appropriate engineering solutions to minimize noise generation. In this study, BSR sounds were classified depending on their periodicity (occurrence per seconds) and spectrum contents (sharpness). Prony's method was used to quantify exponentially decreasing impact noises. The attenuation derived from Prony’s method represented how fast BSR noises decay with time. After performing auditory tests for BSR noises measured in actual vehicles, the rating on the annoyance level was determined. BSR sounds that decayed quickly caused less annoyance. Second, as the study of signal-based fault detect method, The chatter detection method through the vibration signal processed is proposed. Changing cutting condition such as cutting rotating speed and depth, the signal of chatter occurrence is measured by using accelerometer sensor. When chatter appears, the frequency response of vibration signal is composed of harmonic tonal components by spindle rotation and others tonal components. Using this property, index that represents characteristic chatter signal is composed of processed frequency response of the signal with moving average filter, and envelope using Hilbert-transform. And then, We measured the vibration response during normal operation of the microwave machines having wide variety of detects, and sound quality parameters calculated for measured vibration response were applied to analyze the temporal characteristics. Fault diagnosis was performed by using discriminant analysis and support vector machine. The various defects in the system occurring during manufacturing process were accurately identified.
Appears in Collections:
Files in This Item:
There are no files associated with this item.
RIS (EndNote)
XLS (Excel)


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.