Image-Based Analysis of Transformer Oil at Varying Temperatures Using Statistical Hypothesis Testing

Abstract

This paper aims to propose a new method of analysis of transformer oil images by means of statistical tests of hypotheses, such as f-test and t-test. Electrical transformer is continuously operating static electrical equipment. In general, mineral oil extracted from crude petroleum is used as insulation and cooling medium in all power transformers. The performance as well as characteristics of the dielectric oil determines the life of the transformer. In this paper, the real time images of the mineral oil acquired at different temperatures using digital camera. The captured images are preprocessed using image noise eliminating filters such as wiener, non-local means, shock, complex shock and improved complex shock filters. Simple statistical measures such as mean, standard deviation, mode, minimum, maximum, and variance are evaluated for real time as well as for enhanced images. A theoretical analysis presented by performing statistical tests on all images to determine the 0.05 significance level between the means as well as variances. The results of statistical t-test as well as f-test are validated using 95% confidence level. The accurate results are obtained through statistical hypothetical test methods.