کنفرانس مهندسی برق ایران

صفحه اصلی / سی امین کنفرانس بین المللی مهندسی برق

Using a Novel Connection Triangle as a Classifier to Discriminate between Different Faults in the Frequency Response Analysis

نویسندگان :

Mohammad Hamed Samimi¹

1- دانشگاه تهران

کلمات کلیدی :

classification algorithms, fault detection, fault diagnosis, frequency response, power transformers testing

چکیده :

The frequency response analysis (FRA) is an effective method to detect certain mechanical and electrical faults inside transformers. Despite the standard measurement procedure of the frequency response, the interpretation of the results still lacks an objective methodology. Therefore, the interpretation of FRA results is still a subject of study. The interpretation means to detect the existence of a fault, its type, and its extent. As a step forward in this path, this paper suggests a new method to discriminate between different fault types in the FRA method, i.e., the method can detect the fault type. Here, a connection triangle is proposed where the FRA variations corresponding to three different connections are mapped onto a ternary plot. The outcome is that different fault types are accumulated in various areas of the ternary plot, which leads to the fault type classification. In addition to different simulation results, the outcomes of experiments are used in this paper to prove the functionality of the idea.