Multi-Sensor Data Fusion Method for Smart Substations Based on Hierarchical Attention Network and Transfer Learning

The increasing complexity of modern power substations necessitates advanced data fusion techniques for intelligent monitoring, diagnosis, and decision-making. This paper proposes a novel multi-sensor data fusion framework for smart substations, leveraging hierarchical attention networks (HANs) and transfer learning techniques. The HAN effectively captures the multi-level dependencies and importances of different sensors, while transfer learning adapts the pre-trained model to specific substation environments. Extensive experiments on two real-world datasets demonstrate the superiority of the proposed framework over state-of-the-art methods in sensor fault detection and substation state estimation tasks, achieving an average <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{F 1}$</tex>-score of 0.95 and a mean squared error of 0.018, respectively. The framework's practical applicability is showcased through case studies, highlighting its potential for intelligent monitoring, diagnosis, and decision-making in smart substations. Future research directions include incorporating additional contextual information, developing online learning mechanisms, enhancing interpretability, and exploring scalability aspects. The proposed framework represents a significant step towards revolutionizing the operation and maintenance of smart substations, contributing to a more reliable, efficient, and sustainable power grid.