数字孪生建模支撑数据

数据集资源来源于 2023 年 12 月 - 2024 年 1 月国能黄骅港 R5 取料机的实际运行场景，采用多源异构传感器融合与高清视觉采集相结合的方式产生：结构化数据通过部署在回转轴承、驱动机构等关键部位的温振传感器、位移传感器同步采集，经中控系统上传后导出；非结构化视频数据由安装在 R5 大臂皮带关键位置的高清网络枪型摄像机以固定视角、稳定帧率拍摄获取。所有原始数据均经过严格预处理，包括剔除离群值与无效空值的数据清洗、基于时间序列插值算法的数据补全，以及运用小波变换或卡尔曼滤波的数字去噪，确保数据质量可靠。 数据集核心内容包含三类相互关联的多模态数据，共计 2 个 *.csv 格式结构化文件和 5个 *.mp4 格式非结构化视频文件，总数据量 15.06GB。其中，“装船作业流程数据” 记录了序号、流程号、作业日期、工班、作业起止时刻、取料垛位、作业吨数等关键信息，精准描述每次取料装船作业的完整过程；“设备停机记录数据” 涵盖序号、流程号、停机日期、工班、停机设备、停机起止时刻、停机原因编号及具体原因等字段，系统记录设备故障或检修事件；R5 大臂皮带枪机作业视频则直观呈现皮带运行状态、物料流情况及各类视觉可辨的异常现象，与结构化数据在时间线上对齐形成互补验证。This dataset is sourced from the actual operational scenarios of the China Energy Huanghua Port R5 reclaimer between December 2023 and January 2024. It was generated through a combination of multi-source heterogeneous sensor fusion and high-definition visual acquisition. Structured data was synchronously collected by temperature-vibration sensors and displacement sensors deployed at key components including rotary bearings and drive mechanisms, then uploaded and exported via the central control system. Unstructured video data was captured by high-definition network box cameras installed at key positions of the R5 boom belt, with fixed viewing angles and stable frame rates. All raw data underwent rigorous preprocessing, including data cleaning to remove outliers and invalid null values, data imputation via time-series interpolation algorithms, and digital denoising using wavelet transform or Kalman filtering, to ensure reliable data quality. The core content of the dataset comprises three types of interrelated multimodal data, totaling 2 structured files in *.csv format and 5 unstructured video files in *.mp4 format, with a total data volume of 15.06 GB. The "Ship Loading Operation Process Data" records key information such as serial number, process number, operation date, shift, start and end time of operation, reclamation stockpile position, and operation tonnage, accurately describing the complete workflow of each reclamation and ship loading operation. The "Equipment Shutdown Record Data" covers fields including serial number, process number, shutdown date, shift, shutdown equipment, start and end time of shutdown, shutdown cause code and specific cause, systematically documenting equipment faults or maintenance events. The operation videos captured by the box cameras on the R5 boom belt intuitively display the belt operating status, material flow conditions and all visually identifiable abnormal phenomena, which are temporally aligned with the structured data to enable complementary verification.

搜集汇总

背景与挑战

背景概述

该数据集源自2023年12月至2024年1月国能黄骅港R5取料机的实际运行场景，通过传感器与视觉采集结合生成，包含装船作业流程和设备停机记录两类结构化数据，以及大臂皮带作业视频等非结构化数据。数据集经过清洗、补全和去噪预处理，总规模为15.06GB，包括2个CSV文件和5个MP4文件，共计8个文件，旨在支持数字孪生建模应用。