A Fingerprint Recognition Method Based on Deep Learning
Xin Li, Ming Xie, Dongmei Bin · 2024
With the rapid advancement of electric power Internet of Things (IoT) technology, its influence in monitoring, managing, and enhancing the efficiency of power systems has become increasingly significant. This study explores fingerprint recognition technology based on the ResNet (Residual Network) architecture to address the limitations of traditional fingerprint recognition methods within the complex environment of the power IoT. By utilizing the ResNet network, this study successfully overcame the issue of gradient vanishing in deep neural networks, thereby boosting the accuracy of fingerprint recognition to 97.1%. This improvement significantly enhances security verification in the power IoT. Furthermore, the research findings suggest that ongoing advancements in model optimization strategies and the practical deployment of deep learning technologies could further bolster the security and operational efficiency of smart grids. This study not only introduces a novel method for fingerprint recognition but also provides valuable theoretical and practical insights for optimizing the security authentication systems of the power IoT.