Defect detection on electrical power equipment using thermal imaging technology

• Main Author: Geoffrey, Ogadimma Asiegbu
• Format: Thesis
• Language: English
• Published: 2013
• Subjects: TK Electrical engineering. Electronics Nuclear engineering
• Online Access: http://umpir.ump.edu.my/id/eprint/7598/1/Defect%20detection%20on%20electrical%20power%20equipment%20using%20thermal%20imaging%20technology.pdf; http://umpir.ump.edu.my/id/eprint/7598/
• Institution: Universiti Malaysia Pahang

Electrical power equipment and components are vital constituent portions of human existence. They are found virtually in every domestic home and manufacturing industries. These electrical power equipment, operates at a temperature above absolute zero, certainly emit infrared radiation. In some power distribution systems, existing station equipment could no longer withstand the short circuit current capacity causing equipment break down. The electrical equipment failures can be avoided if the temperature threshold is detected in order to take timely corrective action. Quality control and part inspection have done considerably well in the area of manufacturing but not yet gotten to its fully robust thermal imaging technology application. Thermal image is a term comes from the infrared thermography. It has gained its popularity in the last few decades over other predictive maintenance techniques due to its many advantages such as contact-less, easy to interpret the thermal data, large area of inspection as well as free from dangerous radiation. This research project proposed defect detection on electrical power equipment using thermal imaging technology. The aim is to study the thermal characteristic of electrical power equipment, secondly, to design defect detection technique and make a fault decision as well as comparing results with other defect detection techniques and international thermal evaluation standard. A thermal imager is used to acquire thermal images of the tested electrical facilities under various operating conditions. The thermal image in RGB color space is normalized and morphologically dissected using image mean, variance, and covariance which is applied on the mixtures of Gaussian Probability Distribution Function (GPDF) through which threshold values were determined. Using the threshold values similar pixels are connected via maximum likelihood criterion been function of short circuit OR logic operator. The predetermined threshold values are optimized using Receiver Operating Characteristic (ROC) curve and the area under convex hull. Regions of electrical thermogram are segmented using the optimal threshold values. Various features in the infrared thermal (IRT) image are extracted in order to detect anomalies. Classification and decision are made in terms of colors and temperature difference values. Matlab image processing software is used to implement these procedures. Application of this system is quite simple, user friendly, time and cost effective. In conclusion, a total of 111 different electrical power distribution facilities was experimentally inspected. Within the limits of experimental errors, the results of the analysis showed that 99.9% sensitivity and 99.72% accuracy was achieved with an error rate of 0.28 that was attributed to mistakes due to over and less caution during experimental thermal inspection of electrical facilities. The results suggested that, the method provides an accurate identification of defective parts can be extended for further applications. The results also suggest that the system works well enough to help improve the value and efficiency of consumable electrical power equipment reducing the number of faults in the power distribution line, ensuring safety of the workers and users of electricity, protecting electrical power facilities from damage due to over-heating or fire. Above all, testing, inspection and preventive maintenance work become safer, easier, and faster with a reasonably high degree of accuracy with this result-oriented defect detection scrutiny system on electrical power facilities.