Active Noise Control Technology Based on Time Series Forecasting Algorithm for Power Transformer

Due to the accelerated pace of urbanization, some substations are increasingly positioned near residential zones and public spaces. The noise generated by these substations has significantly impaired the work and living conditions of nearby residents, becoming a major source of complaints. Compared to traditional passive noise control technology, active noise control (ANC) technology exhibits significant advantages such as greater controllability, ease of deployment, and enhanced performance in low-frequency noise reduction. The effectiveness of an ANC system primarily depends on the control algorithm. However, traditional least mean square (LMS) algorithms face challenges such as slow convergence and susceptibility to divergence when applied to time-varying systems. Based on the time-domain and frequency-domain characteristics of transformer noise, this paper proposes a control technology utilizing the time series forecasting algorithm and designs an active noise control system for power transformers by integrating both software and hardware parts. The experimental results demonstrate that the system can achieve a maximum overall noise reduction of approximately 8 dB at the error sensor, with the noise reduction for the 100 Hz low-frequency component reaching up to around 25 dB. These results validate the effectiveness of the time series forecasting algorithm and the reliability of the active noise control system for power transformers designed in this study.