基于深度学习的水泥发运码头辅助装船模型研发及应用

Research and application of auxiliary loading model of cement shipping terminal based on deep learning

码头是连接水上运输和陆上运输的重要枢纽。但是由于码头上的装船机工作环境复杂,操作员在装船过程中会面临如无法全面监控作业过程、装料时冒灰、扬尘等问题。为了缓解这种情况,本文提出了一种基于深度学习算法的辅助装船模型,基于先进的YOLOv7和YOLOv8模型,能够实现24 h全天候监控装船过程,实时预测溜筒的位置和高度,对装船时的物料偏移度进行计算,并检测装料时出现的冒灰情况。此外,当溜筒靠近船壁时将发出警告,避免发生碰壁情况。

The dock is an important hub connecting water transportation and land transportation.However,due to the complex working environment of the loading machine on the dock,operators may face many problems during the loading process,such as the inability to fully monitor the operation process,and the emission of ash/dust during loading.In order to alleviate this situation,this paper proposes an assisted ship loading model based on deep learning algorithms.Based on advanced YOLOv7 and YOLOv8 models,it can monitor the loading process 24 hours a day,predict the position and height of the ship loading spout in real time,calculate the material deviation and detect ash emission during loading.In addition,when the ship loading spout approaches the walls of the cargo hold,a warning will be issued to avoid collision.