Scholarlink

<p>Air quality monitoring with low-cost PM<sub>2.5</sub> (particulate matter with aerodynamic size lower than 2.5 microns) particle sensors is becoming increasingly common; nevertheless, the accuracy of these sensors varies among devices and operating conditions. Prior studies have investigated how environmental factors, such as temperature and relative humidity (RH), affect sensor accuracy and have identified significant impacts of these variables in numerous instances. The current study used three relative humidities, 20 ± 3%, 50 ± 3%, and 80 ± 3%, to test and calibrate two low-cost PM<sub>2.5</sub> sensors, the Plantower PMSA003(A) and the Sensirion SPS30, against the standard research-grade monitor Grimm 11-A. Hygroscopic particle NaCl and hydrophobic particle compressor oil are utilized for the purpose of calibration. To process the sensor response, five calibration models: linear, quadratic, piecewise linear, piecewise quadratic, and artificial neural network (ANN) are utilized. To determine the model performance, the coefficient of determination (R<sup>2</sup>), mean bias error (MBE), and root mean square error (RMSE) are calculated. The Sensirion SPS30 outperformed the other sensor (Plantower) in terms of its sensitivity to changes in relative humidity. The ANN model achieved R² values of over 0.9967 and RMSE values below 10% for both types of particles (using the Grimm 11-A as a reference). While sensors’ reactions to relative humidity can vary for hygroscopic particles (NaCl), they show very little fluctuation for hydrophobic particles (compressor oil) across all relative humidity levels. Using the calibration equations available for 20 ± 3%, 50 ± 3%, and 80 ± 3% relative humidity, a weighted calibration method is proposed to determine PM<sub>2.5</sub> levels at any relative humidity ranging from 20 to 80%.</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

Study on calibration of low-cost particulate matter sensors for hydrophilic and hydrophobic particles under varying relative humidity

  • 期刊论文

  • 发表日期：

    2025年11月12日

查看全文
  • Deepali Agrawal,
  • Anil Kumar Saini,
  • Aakash C. Rai,
  • Prateek Kala

摘要

Air quality monitoring with low-cost PM2.5 (particulate matter with aerodynamic size lower than 2.5 microns) particle sensors is becoming increasingly common; nevertheless, the accuracy of these sensors varies among devices and operating conditions. Prior studies have investigated how environmental factors, such as temperature and relative humidity (RH), affect sensor accuracy and have identified significant impacts of these variables in numerous instances. The current study used three relative humidities, 20 ± 3%, 50 ± 3%, and 80 ± 3%, to test and calibrate two low-cost PM2.5 sensors, the Plantower PMSA003(A) and the Sensirion SPS30, against the standard research-grade monitor Grimm 11-A. Hygroscopic particle NaCl and hydrophobic particle compressor oil are utilized for the purpose of calibration. To process the sensor response, five calibration models: linear, quadratic, piecewise linear, piecewise quadratic, and artificial neural network (ANN) are utilized. To determine the model performance, the coefficient of determination (R2), mean bias error (MBE), and root mean square error (RMSE) are calculated. The Sensirion SPS30 outperformed the other sensor (Plantower) in terms of its sensitivity to changes in relative humidity. The ANN model achieved R² values of over 0.9967 and RMSE values below 10% for both types of particles (using the Grimm 11-A as a reference). While sensors’ reactions to relative humidity can vary for hygroscopic particles (NaCl), they show very little fluctuation for hydrophobic particles (compressor oil) across all relative humidity levels. Using the calibration equations available for 20 ± 3%, 50 ± 3%, and 80 ± 3% relative humidity, a weighted calibration method is proposed to determine PM2.5 levels at any relative humidity ranging from 20 to 80%.