Moniri, Farshad,
Volume 2, Issue 1 (January 2006)
Abstract
Power transformers are key components in electrical power supplies and their failure could cause severe consequences on continuity of service and also generates substantial costs. Identifying problems at an early stage, before catastrophic failure occurs, is a great benefit for reliable operation of power transformers. Frequency Response Analysis (FRA) is a new, well-known and powerful diagnostic test technique for transformers which could find mechanical as well as electrical faults such as detection and positioning of winding short circuit, winding movement, loss of clamping pressure, aging of insulation, etc. Yet there are several practical limitations to affect the accuracy and ease using this test as a regular condition monitoring technique in the field that many of them originated from noise and measuring errors. This paper purposes a transformer automated self diagnosis system can be installed on every power supply as a part of SCADA to extract FRA graphs from transformers and offers high repeatability which is a great benefit for FRA test. This is the first time that KALMAN Filter will be use in order to eliminate narrow-band and wide-band noises from FRA graphs that ends up not only smoothed measurement but also rate of changes that is so valuable in decision making and scheduling for transformers maintenance. So we will have an intelligent system which is able to predict the future of transformer using experience of not only own self but also all the transformers in an integrated network.
M. R. Moniri, M. M. Nayebi, A. Sheikhi,
Volume 4, Issue 4 (October 2008)
Abstract
A detector for the case of a radar target with known Doppler and unknown
complex amplitude in complex Gaussian noise with unknown parameters has been derived.
The detector assumes that the noise is an Auto-Regressive (AR) process with Gaussian
autocorrelation function which is a suitable model for ground clutter in most scenarios
involving airborne radars. The detector estimates the unknown parameters by Maximum
Likelihood (ML) estimation for the use in the Generalized Likelihood Ratio Test (GLRT).
By computer simulations, it has been shown that for large data records, this detector is
Constant False Alarm Rate (CFAR) with respect to AR model driving noise variance. Also,
measurements show the detector excellent performance in a practical setting. The detector’s
performance in various simulated and actual conditions and the result of comparison with
Kelly’s GLR and AR-GLR detectors are also presented.
