Grounding Pile Detection System based on Deep Learning

Abstract

The safe and reliable power supply provided by State Grid gives convenience for our life. At the same time, it also plays a central role in the construction and development of country. However, the working environment of State Grid is under high voltage. In order to prevent personal electric shock, damage equipment and lines, prevent fire and lightning, prevent electrostatic damage and ensure the power system operation, the staff must install grounding piles according to the power operation specification. To tackle this problem, this paper proposes a grounding pile detection system based on deep learning network. First, cameras can acquire images of these monitored areas in real time. Then, these images are transmitted to the grounding pile detection system for detection. A warning will be given if it is found that workers have not installed the grounding piles in the monitored areas in accordance with the specifications. At present, there is no research on grounding pile detection. So we created our own dataset. Through experiments, our system achieves 92.00% accuracy, 97.50% accuracy and 13.5% false alarm rate in our dataset.