کنفرانس بین المللی ماشین و محرکه های الکتریکی

صفحه اصلی / The 5th International Conference on Electrical Machines and Drives

A new development in power transformer fault diagnosis using artificial intelligence models, a review study

نویسندگان :

Reza Hojjati¹ Asghar Akbari Azirani²

1- دانشگاه صنعتی خواجه نصیرالدین طوسی 2- دانشگاه صنعتی خواجه نصیرالدین طوسی

کلمات کلیدی :

power transformer،fault diagnosis،condition monitoring،AI،machine learning،deep learning،DGA،FRA،PD

چکیده :

This paper presents a comprehensive review of the application of Artificial Intelligence (AI) algorithms for the fault diagnosis of power transformers, focusing on three principal diagnostic methods, including Dissolved Gas Analysis (DGA), Frequency Response Analysis (FRA), and Partial Discharge (PD) monitoring. For DGA, AI-based methods are shown to enhance diagnostic accuracy and reliability by integrating results from multiple conventional interpretation techniques, generating novel diagnostic features from raw data, and generating new data representations from existing datasets for AI models input. In FRA, AI has demonstrated remarkable effectiveness in identifying the type, location, and severity of mechanical defects by analyzing FRA signatures either through the estimation of transformer ladder model parameters, the use of polar plot representations, or the extraction of statistical indices as model inputs. Finally, for PD monitoring, the application of AI to large, labeled datasets enables high accuracy classification of types of discharge. Notably, AI diagnosis techniques also facilitate the localization of PD events using data from only a single sensor, representing a significant advancement over traditional methods. The findings across these areas confirm that AI provides more robust, precise, and efficient solutions for modern transformer condition monitoring and fault diagnosis.