Deep learning-based damage detection of mining conveyor belt

The mining conveyor belt is an important component of the coal mine belt conveyor, which plays the role of carrying materials and transmitting power. Aiming at the problem that mining conveyor belts are easily damaged under severe working conditions, based on the reclassification and definition of conveyor belt damage types, a special data set for conveyor belt damage was established and a new detection method that can simultaneously detect multiple faults based on improved Yolov3 algorithm was proposed. The EfficientNet was adopted as the backbone feature extraction network instead of Darknet53 in the improved algorithm, comprehensively considers the balance between network depth, width, and image resolution for network scaling to improve the accuracy of the algorithm in limited computing resources. Experiments have proved that the improved algorithm in this paper takes into account both detection speed and detection accuracy. The detection speed can reach 42 FPS, and the mean average precision can reach 97.26%. Compared with the original Yolov3 algorithm, the accuracy is increased by 10.4%, with the speed 45.9%, which provides new ideas and methods for ensuring the safe and stable work of conveyor belts.

Automated summary

This study proposes a new detection method based on the improved Yolov3 algorithm for mining conveyor belt damage, using EfficientNet as the backbone feature extraction network to balance network depth, width, and image resolution, achieving an accuracy of 97.26%. The improved algorithm balances detection speed and accuracy, with a detection speed of 42 FPS and mean average precision of 97.26%.