地铁轨道现场扣件巡检数据

在地铁运营中，轨道扣件的状态直接关系到列车运行的平稳性与安全性。通过工电隧综合巡检机器人采集地铁轨道的扣件巡检图像，结合AI算法分析轨道左右侧的扣件设备状态，判断扣件中弹条或螺栓是否松动、缺失、断裂等状态，实现轨道两侧扣件的状态预警与报警，为列车的安全、稳定运营提供了有力保障。通过实时上传采集地铁轨道的扣件巡检图像，分析轨道左右侧的扣件设备状态，实现轨道两侧扣件的状态预警与报警，具体过程：
（1）数据采集：机器人在以15km/h的运行下，预设线扫相机采样频率、采样间隔、行车方向等信息实时采集轨道两侧的扣件图像；
（2）数据处理：将采集的扣件图像传输至算法识别模块，算法进行深度学习YOLOv8目标检测模型进行扣件中的弹条、轨距块、螺栓等设备的检出。针对弹条进行深度学习Unet++分割网络的处理，判断弹条的完整性，是否存在缺失、松动等状态。针对轨距块同样进行分割网络的处理，判断轨距块是否存在缺失状态。针对螺栓需结合深度信息判断螺栓时候存在松动状态。
（3）将判断分析后的扣件设备（弹条、轨距块、螺栓等）状态值上传至界面进行结果展示，实现扣件设备的预警与报警，1为预警，0为不预警，便于现场人员快速判别设备的状态是否异常；
（4）同时可根据采集所得的数据按天、月、年等时间维度进行趋势分析，形成安全报表数据。

In subway operations, the condition of rail fasteners is directly linked to the smoothness and safety of train operations. Subway track fastener inspection images are collected by integrated inspection robots for railway, power and tunnel maintenance, and then AI algorithms are employed to analyze the status of fastener equipment on both left and right sides of the track, identifying abnormalities such as loose, missing or broken elastic strips and bolts. This enables early warning and alarm for fasteners on both track sides, providing a strong guarantee for the safe and stable operation of trains.

By uploading the collected subway track fastener inspection images in real time and analyzing the status of fastener equipment on both track sides, the aforementioned early warning and alarm functions for track fasteners are realized. The specific workflow is as follows:
1. Data Collection: When traveling at a speed of 15 km/h, the robot collects fastener images on both sides of the track in real time by presetting parameters such as the sampling frequency, sampling interval, and driving direction of the line-scan camera.
2. Data Processing: The collected fastener images are transferred to the algorithm recognition module, which utilizes the deep learning YOLOv8 object detection model to detect key components of fasteners including elastic strips, gauge blocks and bolts. For elastic strips, the deep learning Unet++ segmentation network is applied to assess their integrity and detect anomalies such as missing or loose states. For gauge blocks, the segmentation network is also used to determine whether they are missing. For bolts, depth information is integrated to judge whether they are loose.
3. Result Display and Alarm: The status values of the analyzed fastener components (elastic strips, gauge blocks, bolts, etc.) are uploaded to the interface for result visualization, realizing early warning and alarm for fastener equipment, where 1 indicates an early warning and 0 indicates no early warning, allowing on-site personnel to quickly identify abnormal equipment status.
4. Trend Analysis and Reporting: Additionally, trend analysis can be conducted on the collected data based on time granularities such as day, month and year, generating safety report data.

搜集汇总

背景与挑战

背景概述

该数据集包含501条地铁轨道扣件巡检数据，每日更新，采用xlsx格式。数据通过工电隧综合巡检机器人采集，结合YOLOv8和Unet++等AI算法分析扣件状态，用于预警和报警，保障列车安全运营。