Please use this identifier to cite or link to this item:
http://localhost:8081/xmlui/handle/123456789/1802
Full metadata record
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Ahmad, Khwaja Hina | - |
dc.date.accessioned | 2014-09-25T13:00:24Z | - |
dc.date.available | 2014-09-25T13:00:24Z | - |
dc.date.issued | 2010 | - |
dc.identifier | Ph.D | en_US |
dc.identifier.uri | http://hdl.handle.net/123456789/1802 | - |
dc.guide | Kumar, Vinod | - |
dc.guide | Gupta, S. P. | - |
dc.description.abstract | The present research work is focussed to detect bearing faults in electrical machines at the incipient stage through analysis of the monitoring data. Analysis is done on the vibration signal which is acquired using the accelerometer. Investigations of bearing faults are carried out on the basis of statistical analysis of vibration signal in time domain. The basic vibration data is obtained on a 7.5kW, 3-phase, 415Volts, 50Hz, 7.5kW, 15A, 1440 rpm cage Induction motor. A DC generator, directly coupled to the motor, is used for loading. The vibration signal is picked up by placing the vibration transducer, which is of piezo-electric type (PU-601R) on the bearing cap of the loadend bearing. For signal conditioning, a machine analyzer MK-500 is employed, the output of which is an analog signal which is converted in digital form by NI-6024E card. The data is acquired in the PC using the LabVIEW software and stored in Excel file. The statistical analysis to determine time domain features is done in EXCEL. The vibration signal in terms of acceleration is obtained for a given bearing while the motor is running on selected loading condition. In each run 50,000 data sample points are obtained at a sampling frequency of 1280Hz. For ensuring consistency a number of runs are taken for each operating condition. The bearings used in the experimentation have been obtained directly from the manufacturer - National Engineering Industries Limited, Jaipur, India. A set of healthy bearings (NBC 6308) and bearings having a small point fault in (a) outer race and (b) inner race, supplied by the manufacturer, have been employed in the present work. In machine condition monitoring, time domain analysis is used for studying the time waveform of the vibration signal. To enhance feature extraction, different time domain techniques are used. For a time domain signal, the characteristic features are: peak value, RMS value, crest factor, standard deviation, kurtosis, geometric mean and skewness. Upon onset of a fault, the variation in the values of these characteristic features is examined for diagnosis purpose. The present research work implements the statistical method which can be categorized firstly, into statistical parameters which are: the mean, standard error, median, mode standard deviation, sample variance, kurtosis, skewness, range, minimum, maximum and confidence interval and secondly, the statistical i inference, namely the probability density function (PDF) and cumulative distribution function(CDF). Statistical Analysis is broadly classified into two major components namely 1) Statistical parameters which includes the a) Measure of Central Tendency-(MCT): mean, median and mode b) Measure of Variability-(MV): range, variance, standard deviation and standard error c) Measure of Dispersion-(MD): kurtosis and skewness 2) Statistical Inferences which includes the a) Probability Density Function (PDF) or Gaussian distribution b) Cumulative Distribution Function (CDF) The statistical parameters are calculated using 50000 data point in EXCEL. To determine the Gaussian distribution, the 50000 vibration data points are broken down by Sturge's formula into 17 classes. The FFT is also calculated and plotted to determine the bearing fault frequencies, present in the spectra. This was done for healthy (fault free) bearing, outer race fault and inner race faults. Further, for each of the cases, four load conditions were investigated namely the no load, slight load, half load and three quarter load. In present work some of the findings are: 1) For the statistical parameters the measure of variability is found to show a marked change both in the case of occurrence of a fault as well as when the load is gradually increased. On the other hand no significant change is seen in the measure of central tendency and measure of dispersion for an incipient fault. 2) As the fault progresses the pdf spreads and the amplitude dips. This indicates that the vibration levels have increased as also the range in both the pdf and cdf. 3) The vibration signal at incipient stage of fault remains Gaussian. | en_US |
dc.language.iso | en | en_US |
dc.subject | ELECTRONICS AND COMPUTER ENGINEERING | en_US |
dc.subject | ELECTRICAL MACHINE BEARING FAULTS | en_US |
dc.subject | PROBABILITY DENSITY FUNCTION | en_US |
dc.subject | CUMULATIVE DISTRIBUTION FUNCTION | en_US |
dc.title | ELECTRICAL MACHINE BEARING FAULTS: STATISTICAL ANALYSIS AND DIAGNOSTICS | en_US |
dc.type | Doctoral Thesis | en_US |
dc.accession.number | G21276 | en_US |
Appears in Collections: | DOCTORAL THESES (E & C) |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
ELECTRICAL MACHINE BEARING FAULTS STATISTICAL ANALYSIS AND DIAGNOSTICS.pdf | 11.08 MB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.