Scholarlink

<p>In order to improve the reliability of low-cost particle sensors in urban air quality monitoring, data quality problems caused by field interference and environmental factors are solved. The study deployed a low-cost sensor system near an urban standard monitoring station in June 2024 to monitor PM<sub>2.5</sub>, PM<sub>10</sub> concentration and ambient temperature and humidity data. The study proposed a physics-based calibration model, using the particle size information provided by the sensor to correct the data, and discussed the impact of the number of channels on the physical calibration model. The 6-channel and 32-channel PM₂.₅ sensors showed correlation coefficients of 0.59 with the reference method, while the 16-channel sensor exhibited a slightly higher correlation (R<sup>2</sup> values 0.62). The result shows that the 16-channel data is considered to have a good fit for the physical model. This study compared the physical model with the machine learning model calibration and validated it with new monitoring data from July. In the second monitoring activity, the physical calibration model (R<sup>2</sup> 0.74 (PM<sub>2.5</sub>) and 0.65 (PM<sub>10</sub>)) outperformed the CatBoost model (R<sup>2</sup> 0.57 (PM<sub>2.5</sub>) and 0.53 (PM<sub>10</sub>)). The result shows that the physics-based calibration model shows stronger generalization in the second monitoring, can adapt to different environmental conditions, and maintain a good calibration effect. The research shows that the application of physics-based calibration model in low-cost sensor embedded system significantly improves the accuracy of urban air quality monitoring and provides a stable and sustainable calibration method.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

In-situ Validation of Low-cost Sensors Based on Physical Calibration Model

  • 期刊论文

  • 发表日期：

    2025年7月8日

查看全文
  • Xiaoting Li,
  • Dou Liu,
  • Rui Yang,
  • Ning Xue,
  • Lina Zheng

摘要

In order to improve the reliability of low-cost particle sensors in urban air quality monitoring, data quality problems caused by field interference and environmental factors are solved. The study deployed a low-cost sensor system near an urban standard monitoring station in June 2024 to monitor PM2.5, PM10 concentration and ambient temperature and humidity data. The study proposed a physics-based calibration model, using the particle size information provided by the sensor to correct the data, and discussed the impact of the number of channels on the physical calibration model. The 6-channel and 32-channel PM₂.₅ sensors showed correlation coefficients of 0.59 with the reference method, while the 16-channel sensor exhibited a slightly higher correlation (R2 values 0.62). The result shows that the 16-channel data is considered to have a good fit for the physical model. This study compared the physical model with the machine learning model calibration and validated it with new monitoring data from July. In the second monitoring activity, the physical calibration model (R2 0.74 (PM2.5) and 0.65 (PM10)) outperformed the CatBoost model (R2 0.57 (PM2.5) and 0.53 (PM10)). The result shows that the physics-based calibration model shows stronger generalization in the second monitoring, can adapt to different environmental conditions, and maintain a good calibration effect. The research shows that the application of physics-based calibration model in low-cost sensor embedded system significantly improves the accuracy of urban air quality monitoring and provides a stable and sustainable calibration method.