Deep-Learning-Based Thickness Detection Method of Ice Covering

Accurate and timely monitoring of icing is of great significance for improving the power grid's disaster prevention and mitigation capabilities and ensuring the safe operation of the power grid. However, the current icing thickness monitoring technology cannot meet actual application requirements in terms of automation and accuracy. As an emerging artificial intelligence technology, deep learning algorithms have been widely used in the field of computer vision. Aiming at the shortcomings of the existing methods for icing thickness recognition, this paper proposes an intelligent recognition method for power grid icing thickness based on the Mask R-CNN model in deep learning. First, image enhancement of the sample data is performed based on the deep learning image enhancement technology, and then the icing thickness sample image database is constructed, and finally the icing thickness recognition is performed based on the Mask R-CNN model. Experimental results show that the method proposed in this paper can achieve high recognition accuracy for all types of icing grades. The quantitative analysis results show that the mAP, mAP50 and mAP75 of the model recognition results are 0.634, 0.966 and 0.779, respectively. The icing thickness identification method based on deep learning proposed in this paper can effectively monitor the icing of power grid equipment, and has an important theoretical guiding role for power grid disaster prevention and mitigation.