Sound signal-based transformer operation status monitoring system

• Published in: 2023 4th International Conference on Computer Engineering and Application (ICCEA) pp. 375 - 380
• Main Authors: Xu, Haifeng, Cai, Jidong, Xu, Long, Xu, Fei
• Format: Conference Proceeding
• Language: English
• Published: IEEE 07-04-2023
• Subjects: adaptive noise cancelling; Adaptive systems; Classification algorithms; Feature extraction; machine health monitoring; machine learning; Machine learning algorithms; Noise cancellation; Noise reduction; random forest algorithm; Transformers
• ISSN: 2159-1288
• DOI: 10.1109/ICCEA58433.2023.10135233

This paper proposes a system for monitoring the operational status of transformers based on sound signals. The system employs an adaptive noise reduction technique to preprocess the sound signals and a random forest algorithm to classify the signals into different operational states. The adaptive noise reduction technique effectively removes background noise and other irrelevant sound signals, allowing for more accurate monitoring of the transformers. The random forest algorithm is a popular machine learning technique that can handle complex, high-dimensional datasets, making it well-suited for analyzing the large number of features extracted from sound signals. Experimental results from field tests on four transformers demonstrate that the proposed algorithm is effective in identifying the operational states of the transformers and can provide timely alerts when faults occur. The system presented in this paper can serve as a useful tool for monitoring the health of transformers.