This paper proposes a method of automatically detecting and classifying low frequency\nnoise generated by power transformers using sensors and dedicated machine learning algorithms.\nThe method applies the frequency spectra of sound pressure levels generated during operation by\ntransformers in a real environment. The spectra frequency interval and its resolution are automatically\noptimized for the selected machine learning algorithm. Various machine learning algorithms,\noptimization techniques, and transformer types were researched: two indoor type transformers from\nSchneider Electric and two overhead type transformers manufactured by ABB. As a result, a method\nwas proposed that provides a way in which inspections of working transformers (from background)\nand their type can be performed with an accuracy of over 97%, based on the generated low-frequency\nnoise. The application of the proposed preprocessing stage increased the accuracy of this method by\n10%. Additionally, machine learning algorithms were selected which offer robust solutions (with the\nhighest accuracy) for noise classification.
Loading....