山东电力高等专科学校学报

为实现主变压器跳闸送电后内部异常放电隐患的快速精准排查，提出一种基于多源数据融合与深度学习的智能排查与预警系统。通过部署特高频传感器、声学传感器及温度传感器，实时采集跳闸送电后主变压器内部的声学、电磁及温度信号，利用小波变换提取时域和频域特征，并构建混合深度学习模型，实现异常放电信号的识别与分类。测试结果表明，该系统可实现主变压器跳闸送电后内部异常放电隐患的智能排查与预警，提升电网运行的安全性和可靠性。

To enable rapid and precise detection of abnormal discharge hidden danger in main transformer after trip and energization,an intelligent detection and early-warning system based on multi-source data fusion and deep learning is proposed. The deployment of UHF sensors,acoustic sensors,and temperature sensors facilitates real-time acquisition of acoustic,electromagnetic,and thermal signals in main transformer after trip and energization. The wavelet transform is used to extract the time-domain and frequency-domain features,and a hybrid deep learning model is constructed to achieve the recognition and classification of anomalous discharge signals. Test results confirm that the system can achieve intelligent detection and early-warning of abnormal discharge hidden danger in main transformer after trip and energization,enhancing the safety and reliability of power grid operation.