Power transformers are vital in power systems, where oil temperature is a key operational indicator. This study proposes an advanced hybrid neural network model, BWO-TCN-BiGRU-Attention, to predict the top-oil temperature of transformers. The model was validated using temperature data from power transformers in two Chinese regions. It achieved MAEs of 0.5258 and 0.9995, MAPEs of 2.75% and 2.73%, and RMSEs of 0.6353 and 1.2158, significantly outperforming mainstream methods like ELM, PSO-SVR, Informer, CNN-BiLSTM-Attention, and CNN-GRU-Attention. In tests conducted in spring, summer, autumn, and winter, the model’s MAPE was 2.75%, 3.44%, 3.93%, and 2.46% for Transformer 1, and 2.73%, 2.78%, 3.07%, and 2.05% for Transformer 2, respectively. These results indicate that the model can maintain low prediction errors even with significant seasonal temperature variations. In terms of time granularity, the model performed well at both 1 h and 15 min intervals: for Transformer 1, MAPE was 2.75% at 1 h granularity and 2.98% at 15 min granularity; for Transformer 2, MAPE was 2.73% at 1 h granularity and further reduced to 2.16% at 15 min granularity. This shows that the model can adapt to different seasons and maintain good prediction performance with high-frequency data, providing reliable technical support for the safe and stable operation of power systems.

Fault diagnosis and prediction are critical for ensuring the reliability and efficiency of power equipment in modern electrical grids. Traditional diagnostic methods, including rule-based systems and machine learning approaches, often struggle with high-dimensional data and nonlinearity, limiting their effectiveness in real-world scenarios. This study proposes a deep learning-based multi-modal fault diagnosis and prediction model that integrates sensor readings, historical fault records, and environmental conditions. The model leverages a hybrid CNN-LSTM-Attention architecture, where convolutional neural networks (CNN) extract spatial dependencies, long short-term memory (LSTM) networks model sequential patterns, and an attention mechanism enhances feature selection. The proposed method is evaluated on a real-world power grid dataset, demonstrating superior fault detection accuracy $(96.5 \%)$, extended fault prediction lead time (12.4 hours), and reduced false alarm rates compared to traditional methods. The findings suggest that deep learningbased multi-modal data fusion significantly improves fault diagnosis and predictive maintenance in power systems, offering a scalable and effective solution for intelligent grid management. Future work will focus on improving model interpretability and efficiency for real-time deployment in industrial applications.