基于注意力机制的轻量级矿井钢丝绳断丝检测算法研究

Research on broken wire detection algorithm of lightweight mine wire rope based on attention mechanism

立井提升系统作为煤矿生产中的主要运输设备,其核心构件钢丝绳常因工作负荷大、受到腐蚀、磨损等原因而产生断丝引发事故。传统的立井提升机钢丝绳检测算法存在效率低、劳动强度大、智能化程度差和准确率低等问题。基于此,提出一种改进YOLOv5s模型,并基于改进的模型进行矿井钢丝绳断丝检测。首先,设计Swiener滤波算法进行钢丝绳图像运动模糊修复,抑制噪声干扰;其次,在特征提取阶段,引入RFC3轻量化模块降低模型可训练参数,提升钢丝绳检测速度;第三,提出CBAM R注意力机制,增强模型对小断口断丝的检测能力;最后,引入Focal EIoU损失函数,提高模型对小断口断丝的检测精度并加速模型收敛。研究结果表明:所提出的基于注意力机制矿用钢丝绳断丝检测算法(CTR YOLO)可以更好地满足实际应用需求,减少了误检、漏检导致的人力成本浪费及安全事故的发生。

The shaft lifting system is the primary transportation equipment in coal mine production,wire rope,as the core component,often experiences breakage due to heavy workloads,corrosion,wear,and other factors,leading to accidents.The traditional wire rope detection algorithm for shaft hoists has limitations such as low efficiency,high labor intensity,poor intelligence,and low accuracy.Therefore,the authors proposed an improved YOLOv5s model for detecting broken wires in mine wire ropes.A Swiener filter algorithm was designed to repair motion blur in wire rope images and suppress noise interference;in the feature extraction stage,the RFC3 lightweight module was introduced to reduce trainable parameters and increase detection speed;a CBAM R attention mechanism was proposed to enhance the detection ability for small fracture;the introduction of Focal EIoU loss function improved small fracture detection accuracy and accelerated model convergence.Experimental results demonstrated that the broken wire detection algorithm for mine wire rope based on attention mechanism(CTR YOLO)better met practical application requirements by reducing labor costs waste and minimizing safety accidents caused by false detection or missed detection.