Evaluatioin And Model Calibration For Light Scattering Particulate Matter Sensors

In recent years,the refined treatment of air pollution has been gradually deepened,and higher requirements have been put forward for air quality monitoring and pollution monitoring and traceability.Low-cost sensor technology can be flexibly applied in different scenarios,as a supplement to traditional monitoring methods which can realize high spatial and temporal resolution monitoring.However,as it is an emerging monitoring technology,the accuracy and reliability of measurement data in different application scenarios still need a lot of demonstration,and there is a lack of research on model calibration for sensors in different application scenarios.In this paper,the monitoring data of 9 air automatic monitoring standard stations in Jinan were taken as the standard,the performance of particulate sensors under different application scenarios was evaluated,the influence of environmental factors on the performance of sensors was analyzed,a two-step calibration model was designed and typical application cases of mobile sensors were analyzed.The main conclusions are as follows:In the comparison with standard instruments,the fixed sensor showed better performance that the R2 was 0.89-0.96(PM2.5)0.75-0.89(PM10),while the mobile sensor was not satisfactory,which the R2 was only 0.70-0.88(PM2.5),0.51-0.81(PM10).The absolute error in the mobile sensor is also greater than the fixed sensor.And the mobile sensors equipped on taxis are in the road environment,resulting in the larger error and the worse accuracy.At the same time,the performance of both fixed sensor and mobile sensor in PM2.5 measurement was better than PM10.Relative humidity had a more significant influence on the PM2.5 monitored by the sensor.The correlation between the sensor and the standard instrument became worse with the increase of RH.At high RH(RH>80%),the monitored value of the sensor was higher than that of the standard instrument.PM10 monitored by the sensor was more significantly affected by particle size distribution(PM2.5/PM10).As the PM2.5/PM10 ratio increased,the correlation between the sensor and the standard instrument showed a trend of increasing first and then decreasing,and the best correlation was obtained when the PM2.5/PM10 ratio was between 0.5 and 0.6.The two-step calibration model was superior to the single model and can effectively improve the quality of sensor data.In comparison with standard instruments,the LR-final model increased the R2 values of the PM2.5 and PM10 measured by fixed sensors from 0.89 and 0.79 to 0.98 and 0.97,respectively.The R2 values of PM2.5 and PM10 measured by the mobile sensors were both increased to 0.99 from 0.79 and 0.62.In the application of the two-step model,to ensure that the uncertainty is less than 50%,the concentration of PM2.5 and PM10 of the fixed sensor should be greater than 30μg/m3 and 50μg/m3,respectively;the concentration of PM2.5 and PM10 of the mobile sensor should both be greater than 20μg/m3.In the case of urban road dust pollution characteristics,the heavily polluted areas of Jinan were not located in the densely populated city centre,but concentrated in the sparsely populated suburbs.The road network system in Jinan was divided into 1021 sections,of which 65%the PM2.5 concentration was concentrated at 43μg/m3-46μg/m3,and PM10 concentration was concentrated at 55μg/m3-70μg/m3.Compared with the urban environment(monitoring stations),the pollution of road environment(sensors)in Jinan was seriously affected by the peak time,especially the morning peak time.The sensor monitoring signal was divided into regional pollution signals and emission pollution signals based on the extracted PM baseline and the average value per hour of sensors.During the study,for PM2.5,the regional pollution and emission ratio was 78.6%and 21.4%respectively and that was 71.9%,and 28.1%respectively for PM10.