装配式多功能智能路面板感知数据集

数据集2023YFE0202400-006 装配式多功能智能路面板感知数据集为自主采集数据，在质量控制中计划实施多维度验证机制。智能路面板传感器信号原始测试数据采用100%全字段人工复核，对每一条传感器感知信号数据进行查验，确保其感知信号记录时段对测试车辆加载时段的覆盖率达到100%。以车载摄像头录像与路侧摄像头录像双重校核车辆经过智能路面板的时间节点，以此时间节点为时间中点，向前后各截取1.5s的Z轴加速度信号，获得11传感器、12传感器、21传感器、22传感器、31传感器、32传感器6只传感器，在测试车辆碾压经过路面板前后3s内产生的Z轴加速度感知信号，通过双人交叉校验的方式确保信号截取时间段的正确性。采用两阶段校验法对智能路面板正式运营感知数据的正确性与可靠性进行校验，一阶段采用算法自动校验100%数据样本的时间序列合理性，二阶段通过人工复核10%数据样本的正确性与可靠性，以人工核验数据样本通过率达到100%为标准，确保数据集质量。

Dataset 2023YFE0202400-006: Prefabricated Multifunctional Intelligent Pavement Slab Perception Dataset is self-collected, and a multi-dimensional verification mechanism is planned for quality control. The raw test data of sensor signals from the intelligent pavement slab undergoes 100% full-field manual review; each piece of sensor perception signal data is inspected to ensure that the coverage rate of the perception signal recording period over the loading period of the test vehicle reaches 100%. The time node when the vehicle passes the intelligent pavement slab is verified by both on-board camera and roadside camera recordings. Taking this time node as the midpoint, 1.5 seconds of Z-axis acceleration signals are extracted before and after this point, obtaining the Z-axis acceleration perception signals generated by 6 sensors (11,12,21,22,31,32) within 3 seconds before and after the test vehicle rolls over the pavement slab. The correctness of the signal extraction time period is ensured through double cross-verification. A two-stage verification method is used to verify the correctness and reliability of the formal operation perception data of the intelligent pavement slab. In the first stage, an algorithm is used to automatically verify the rationality of the time series for 100% of the data samples. In the second stage, 10% of the data samples are manually reviewed for correctness and reliability. The dataset quality is ensured by the standard that the pass rate of manually verified data samples reaches 100%.

搜集汇总

背景与挑战

背景概述

该数据集为自主采集的智能路面板传感器信号原始测试数据，通过多维度验证机制确保数据质量，涵盖6只传感器在车辆碾压前后3秒内的Z轴加速度感知信号。数据集总大小为10.75GB，包含818个文件，格式包括xlsx、tar、csv和docx。