Matter Sensor Research Articles

Benchmarking Low-Cost Particulate Matter Sensors: Evaluating Performance Under Controlled Environmental Conditions Using Low-Cost Experimental Setups

Particulate matter (PM) is widely recognized as a major air pollutant with significant impacts on human health, highlighting the need for accurate monitoring. In developing countries, low-cost sensors are crucial for accessible PM monitoring, but their accuracy and reliability must first be assessed. This study benchmarked the Alphasense OPC-N3 and Next PM sensors through laboratory and field evaluations. Laboratory tests were performed in controlled conditions with HEPA-filtered air at low humidity and varying concentrations of water droplets from nebulized deionized water. A 27-day field study in Cienfuegos, Cuba, provided additional insights into real-world performance. The OPC-N3 showed susceptibility to perturbations and was more affected by outliers (especially PM10), relative humidity, and interference from aqueous aerosols. In contrast, the Next PM sensor demonstrated superior stability, lower noise levels, and consistent performance across different environmental conditions. Despite a substantial price difference, both sensors provided valid measurements. Additionally, both sensors produced lognormal PM concentration distributions during field campaigns. This feature could aid in addressing the calibration stability challenges commonly associated with low-cost sensors through in situ calibration methods. While the PM measurements by affordable sensors are not perfect, they are sufficiently reliable for supporting air quality assessments in resource-limited settings.

In-situ validation of embedded physics-based calibration in low-cost particulate matter sensor for urban air quality monitoring

In-situ validation of embedded physics-based calibration in low-cost particulate matter sensor for urban air quality monitoring

Using low-cost sensors to assess common air pollution sources across multiple residences

The rapid development of low-cost sensors provides the opportunity to greatly advance the scope and extent of monitoring of indoor air pollution. In this study, calibrated particle matter (PM) sensors and a non-negative matrix factorisation (NMF) source apportionment technique are used to investigate PM concentrations and source contributions across three households in an urban residential area. The NMF is applied to combined data from all houses to generate source profiles that can be used to understand how PM source characteristics are similar or differ between different households in the same urban area. PM2.5 and PM10 concentrations in all three houses were greater, more variable, and significantly different to ambient concentrations recorded at a nearby ambient monitoring site. Concentrations were also significantly different between houses, with the World Health Organisation 24-h guideline limits for PM2.5 breached in one household. The applied methodology was highly successful at modelling concentrations for all the houses (R2≥ 0.983), finding that across the houses the I/O (indoor to outdoor sources ratio) was the lowest for PM1 (down to 0.08), and greatest for PM10 (up to 4.93). Whilst the sources could not be clearly distinguished further than being outdoors or indoors, the methodology provides clear insights to source variability within and between the monitored houses. These results highlight the importance of monitoring indoor air pollution to improve pollution exposure estimates, as whilst people may live in areas with acceptable ambient air quality, they can be exposed to unhealthy concentrations in their own homes. This method may be applied in future studies for extended periods to investigate the influence of source seasonality on PM concentrations or scaled up to investigate source variability across larger geographical areas.

Precision and Accuracy Analysis of PM2.5 Light-Scattering Sensor: Field and Laboratory Experiments

The widely used low-cost particulate matter (PM) sensors in Thailand, such as the DustBoy, require performance improvements to ensure their data align with the established standards set by the US Environmental Protection Agency (US EPA). This study evaluates the accuracy and reliability of the DustBoy, a commonly used PM2.5 monitoring device in Thailand. A comparative analysis was conducted between the DustBoy and the US EPA’s Federal Reference Method (FRM) and Federal Equivalent Method (FEM). The research involved both laboratory and field testing, where the DustBoy’s performance was analyzed at various particulate matter concentration levels and environmental conditions. The study demonstrated that the DustBoy readings diverged from those of standard monitors at higher PM2.5 concentrations; however, a positive correlation between the devices remained evident. Below 100 µg/m3, the DustBoy overestimated PM concentrations compared to the FRM devices but underestimated them compared to the FEM devices. At higher concentrations, the DustBoy showed a significant overestimation, although the data trends aligned with those of standard devices. The sensor performance was also affected by factors such as the sensor age and device model. Corrections were developed to adjust the DustBoy readings to match the reference devices more closely, enhancing the accuracy post-adjustment. These corrections will refine the DustBoy’s public data reporting and serve as guidelines for other low-cost sensors in Thailand.

Improving PM10 sensor accuracy in urban areas through calibration in Timișoara

Low-cost particulate matter sensors (LCS) are vital for improving the spatial and temporal resolution of air quality data, supplementing sparsely placed official monitoring stations. Despite their benefits, LCS readings can be biased due to the physical properties of aerosol particles and device limitations. An optimization model is essential to enhance LCS data accuracy. This paper presents a calibration study of the LCS network of Timișoara, Romania. The calibration began by selecting LCS devices near National Air Quality Monitoring Network (NAQMN) stations and developing parametric models, choosing the best for broader application. Plantower, Sensirion, and Honeywell sensors showed comparable accuracy. Calibration involved clusters within a 750 m radius around NAQMN stations. Models incorporating RH corrections and multiple linear regression (MLR) were fitted. The best model was validated against data from unseen sensors, leading to mean bias errors (MBE) within 9-17% and RMSEs of 33-35%, within sensor uncertainty margins. Applied to the city-wide LCS network, the model identified several stations regularly exceeding the EU daily PM10 threshold, unnoticed by NAQMN stations due to their limited coverage. The study highlights the necessity of granular monitoring to accurately capture urban air quality variations.

Expanding seawater carbon dioxide and methane measuring capabilities with a Seaglider

Abstract. Warming, ocean acidification, and deoxygenation are increasingly putting pressure on marine ecosystems. At the same time, thawing permafrost and decomposing hydrates in Arctic shelf seas may release large amounts of methane (CH4) into the water column, which could accelerate local ocean acidification and contribute to climate change. The key parameters to observing and understanding these complex processes and feedback mechanisms are vastly undersampled throughout the oceans. We developed carbon dioxide (CO2) and CH4 gliders, including standard operational procedures, with the goal that CO2 and CH4 measurements will become more common for glider operations. The Seagliders with integrated Contros HydroC CO2 or CH4 sensors also include conductivity, temperature, depth, oxygen, chlorophyll a, backscatter, and fluorescent dissolved organic matter sensors. Communication via satellite allows for near-real-time data transmission, sensor adjustments, and adaptive sampling. Several sea trials with the CO2 Seaglider in the Gulf of Alaska and data evaluation with discrete water and underway samples suggest nearly “weather-quality” CO2 data as defined by the Global Ocean Acidification Network. A winter mission in Resurrection Bay, Alaska, provided the first insights into the water column inorganic carbon dynamics during this otherwise undersampled season. The CH4 Seaglider passed its flight trials in Resurrection Bay but needs to be tested during a field mission in an area with CH4 concentrations beyond background noise. Both sensing systems are available to the science community through the industry partners (Advanced Offshore Operations and -4H-JENA engineering GmbH) of this project.

Application of a Discrete AlGaN/GaN HEMT Sensor for In‐Situ Regenerative Soot Particulate Detection

AbstractAn in‐situ regenerative soot particulate matter sensor for automotive exhaust systems based on a discrete AlGaN/GaN high electron mobility transistor (HEMT) was developed and tested, where charge‐sensitive AlGaN/GaN HEMT was utilized to detect the signal variations caused by the presence of soot particulate matters, and the micro‐hotplate serves as soot deposition, while the self‐heating unit was used for sensing and in‐situ regeneration. The two parts were connected for electrical signal transmission. Soot particulate were generated by a laminar diesel flame with an adsorption rate of 23.29 μg cm−2 min−1 on the sensor surface, resulting in 1.32% sensing response after 20 s deposition and reached saturation after a 12 minute deposition of soot. After soot deposition, the sensor was successfully regenerated when a voltage of 2 V was applied to the micro‐hot plate, causing in‐situ heat‐up to reach 600°C for 30 seconds. Moreover, benefiting from the advantage of discrete structure, the sensor exhibited consistent and stable response characteristics even after undergoing more than 10 consecutive regeneration cycles, and the replaceable micro‐hotplate could further extend the service life of the sensor. These results show that the discrete AlGaN/GaN HEMT sensor can be a promising soot particulate sensing platform for automotive exhaust systems.

Investigation of the potential of in-line particle concentration measurements in gas particle separation processes by using low-cost particulate matter sensors

In addition to the costly laboratory-based particle measurement devices, low-cost particulate matter sensors for ambient air quality measurements can also be employed for the detection of PM in gas cleaning devices. The ability to spatially resolve the actual level of particle number concentration with high temporal resolution within a gas-particle separation process is of significant value for process monitoring and control. The quantitative reliability of the measurement data from low-cost particulate matter sensors remains uncertain when the particle concentrations in the apparatus exceed the upper measurement limit of the sensors, whether temporarily or permanently. This study examines the potential of low-cost particulate matter sensors for in-line particle concentration measurements in gas-particle separation processes. The measurement performance of the low-cost particulate matter sensors is initially examined under idealised conditions in a dust chamber. This is done in order to quantify the deviations of the low-cost particulate matter sensors in comparison to a state-of-the-art precision laboratory high-cost sensor. Subsequently, both sensor types are installed at a wet separator test facility, following the quantification of their respective measurement performances. The low-cost particulate matter sensors are capable of measuring particle number concentration above their specified limit (up to ≈6.1×104cm−3), depending on the measurement conditions. By applying a 3rd polynomial correction function, obtained from the initial quantification measurement, the low-cost particulate matter sensors are able to reproduce the particle number concentration measured by the high-cost sensor. Furthermore, the low-cost particulate matter sensors are capable of reproducing short-term fluctuations in the particle number concentration.

Structural simulation and performance evaluation of self-priming electrostatic diesel vehicle emission particle sensor

Here is reported a self-priming electrostatic particulate matter (PM) sensor for real-time monitoring of diesel exhaust under different operating conditions. Initially, a multi-physics coupling model for the sensor was established. By varying the exhaust flow and temperature, the emissions of diesel vehicles were simulated to evaluate the sensor. The results showed that within the typical velocities (5 to 60 m/s) and temperature ranges (100 to 500 °C) of diesel exhaust, the sensor outlet pressure was consistently below the inlet pressure, with a maximum differential pressure of 259 Pa, indicating the capability for self-priming sampling of exhaust under different operating conditions. Additionally, the calibration platform was constructed using a miniature inverted soot generator which produced a series of samples to calibrate the relationship between current and mass concentration with a correlation coefficient of 0.99. From 0 to 100 mg/m3, the PM sensor exhibited a relative error between −6.00% and 6.00% with good stability and consistency for samples of different sizes and concentrations, with correlation coefficients exceeding 0.98. The results between the self-developed sensor and a commercial instrument revealed that the former provided high-precision real-time measurement of diesel exhaust.

Simulation of electrostatic particulate matter sensor regeneration based on the particulate deposition patterns

PurposeThis study aims to establish a multi-physics-coupled model for an electrostatic particulate matter (PM) sensor. The focus lies on investigating the deposition patterns of particles within the sensor and the variation in the regeneration temperature field.Design/methodology/approachComputational simulations were initially conducted to analyse the distribution of particles under different temperature and airflow conditions. The study investigates how particles deposit within the sensor and explores methods to expedite the combustion of deposited particles for subsequent measurements.FindingsThe results indicate that a significant portion of the particles, approximately 61.8% of the total deposited particles, accumulates on the inside of the protective cover. To facilitate rapid combustion of these deposited particles, a ceramic heater was embedded within the metal shielding layer and tightly integrated with the high-voltage electrode. Silicon nitride ceramic, selected for its high strength, elevated temperature stability and excellent thermal conductivity, enables a relatively fast heating rate, ensuring a uniform temperature field distribution. Applying 27 W power to the silicon nitride heater rapidly raises the gas flow region's temperature within the sensor head to achieve a high-temperature regeneration state. Computational results demonstrate that within 200 s of heater operation, the sensor's internal temperature can exceed 600 °C, effectively ensuring thorough combustion of the deposited particles.Originality/valueThis study presents a novel approach to address the challenges associated with particle deposition in electrostatic PM sensors. By integrating a ceramic heater with specific material properties, the study proposes an effective method to expedite particle combustion for enhanced sensor performance.

Feasibility Study on the Use of NO2 and PM2.5 Sensors for Exposure Assessment and Indoor Source Apportionment at Fixed Locations.

Recent advances in sensor technology for air pollution monitoring open new possibilities in the field of environmental epidemiology. The low spatial resolution of fixed outdoor measurement stations and modelling uncertainties currently limit the understanding of personal exposure. In this context, air quality sensor systems (AQSSs) offer significant potential to enhance personal exposure assessment. A pilot study was conducted to investigate the feasibility of the NO2 sensor model B43F and the particulate matter (PM) sensor model OPC-R1, both from Alphasense (UK), for use in epidemiological studies. Seven patients with chronic obstructive pulmonary disease (COPD) or asthma had built-for-purpose sensor systems placed inside and outside of their homes at fixed locations for one month. Participants documented their indoor activities, presence in the house, window status, and symptom severity and performed a peak expiratory flow test. The potential inhaled doses of PM2.5 and NO2 were calculated using different data sources such as outdoor data from air quality monitoring stations, indoor data from AQSSs, and generic inhalation rates (IR) or activity-specific IR. Moreover, the relation between indoor and outdoor air quality obtained with AQSSs, an indoor source apportionment study, and an evaluation of the suitability of the AQSS data for studying the relationship between air quality and health were investigated. The results highlight the value of the sensor data and the importance of monitoring indoor air quality and activity patterns to avoid exposure misclassification. The use of AQSSs at fixed locations shows promise for larger-scale and/or long-term epidemiological studies.

In chamber calibration and performance evaluation of air quality low-cost sensors

Assessing individual exposure to PM2.5 (particulate matter of aerodynamic diameter lesser than 2.5 μm) requires precise monitoring of PM2.5 concentrations at specific geographical and temporal scales. This demand is met globally by low-cost particulate matter sensors, although calibrating them is difficult. In this study, four low-cost PM sensors, Sharp GP2Y1010AU0F, Honeywell HPMA115S0-XXX, Plantower PMSA003-A, and Sensirion SPS30, were calibrated and tested using various aerosols. The calibration method has three steps: individual (considering each sensor independently to a single aerosol type; n = 1), combined (all sensors for a specific model together for a specific aerosol type; n = 4), and generic (all sensors for a given model together to all aerosols; n = 16). Sensor responses are processed using linear, quadratic, power-law, and artificial neural network (ANN) algorithms in each calibration stage. Performance metrics, including coefficient of determination (R2), mean absolute percentage error (MAPE), root mean square error (RMSE), and percentage coefficient of variation (% CV), were utilized for assessment. Amongst all the four tested sensors, the Sensirion SPS30 sensors gave the best performance with a minimum R2 value of 0.911 when calibrated with a generic ANN calibration algorithm. Also, the MAPE was less than 10 %, and the RMSE was less than 7 % when exposed to different particles. Sensirion SPS30 showed the lowest inter-sensor variability with % CV less than 6 %. Sensors identified monodisperse polystyrene latex (PSL) particle size in the investigation. Regardless of exposure to 0.3, 0.46, 0.60, or 1.0 μm PSL, the reported number size distribution for the PMSA003 sensor remained consistent and did not align with the results from Grimm. As the PSL size rose, the SPS30 size distribution changed towards larger particle sizes, although it did not always match Grimm data. As the PSL size increased, the sensor's PM1, PM2.5, and PM10 mass proportions altered.

Innovative Air-Preconditioning Method for Accurate Particulate Matter Sensing in Humid Environments.

Smart cities rely on a network of sensors to gather real-time data on various environmental factors, including air quality. This paper addresses the challenges of improving the accuracy of low-cost particulate matter sensors (LCPMSs) which can be compromised by environmental conditions, such as high humidity, which is common in many urban areas. Such weather conditions often lead to the overestimation of particle counts due to hygroscopic particle growth, resulting in a potential public concern, although most of the detected particles consist of just water. The paper presents an innovative design for an indicative air-quality measuring station that integrates the particulate matter sensor with a preconditioning subsystem designed to mitigate the impact of humidity. The preconditioning subsystem works by heating the incoming air, effectively reducing the relative humidity and preventing the hygroscopic growth of particles before they reach the sensor. To validate the effectiveness of this approach, parallel measurements were conducted using both preconditioned and non-preconditioned sensors over a period of 19 weeks. The data were analyzed to compare the performance of the sensors in terms of accuracy for PM1, PM2.5, and PM10 particles. The results demonstrated a significant improvement in measurement accuracy for the preconditioned sensor, especially in environments with high relative humidity. When the conditions were too severe and both sensors started measuring incorrect values, the preconditioned sensor-measured values were closer to the actual values. Also, the period of measuring incorrect values was shorter with the preconditioned sensor. The results suggest that the implementation of air preconditioning subsystems in LCPMSs deployed in smart cities can provide a cost-effective solution to overcome humidity-related inaccuracies, thereby improving the overall quality of measured air pollution data.

Quantitative evaluation of nicotine and particulate matter exposure for passive- and non-smokers via a nicotine passive sampler and a particulate matter sensor

AbstractPassive smoking poses a major hazard to the health of non-smokers. Cigarette combustion emits environmental tobacco smoke (ETS) that contains various pollutants, including volatile organic compounds (VOCs), polycyclic aromatic hydrocarbons, and particulate matter with particles < 2.5 μm in aerodynamic diameter (PM2.5). Therefore, to assess the risk of passive smoking, it is necessary to determine the exposure levels to ETS compounds caused by passive smoking. However, it is difficult to assess exposure to ETS compounds by excluding the effects of other sources of pollution. In this study, we assessed the individual exposure levels to ETS compounds using a nicotine passive sampler and PM2.5 sensor in 258 non-smokers. Median nicotine and PM2.5 concentrations were 0.065 and 7.5 µg m− 3, respectively. The median nicotine concentration was nearly identical to that found in non-smoking areas (0.05 µg m− 3). The median PM2.5 concentrations were below the standard value for indoor PM2.5 concentrations (12 µg m− 3). These results indicated that the effect of passive smoking was rather modest. Nicotine concentrations in individuals exposed to passive smoking indoors were significantly higher than in those not exposed to passive smoking. Nicotine exposure from domestic passive smoking was twice that from non-domestic passive smoking. There was no significant difference in PM2.5 exposure between passive smoking in domestic and non-domestic settings. Moreover, it was suggested that passive smoking was equivalent to the act of smoking 2.4 × 10− 3 cigarettes per day.

DATIV-Remote Enhancement of Smart Aerosol Measurement System Using Raspberry Pi-Based Distributed Sensors.

Enclosed public spaces are hotspots for airborne disease transmission. To measure and maintain indoor air quality in terms of airborne transmission, an open source, low cost and distributed array of particulate matter sensors was developed and named Dynamic Aerosol Transport for Indoor Ventilation, or DATIV, system. This system can use multiple particulate matter sensors (PMSs) simultaneously and can be remotely controlled using a Raspberry Pi-based operating system. The data acquisition system can be easily operated using the GUI within any common browser installed on a remote device such as a PC or smartphone with a corresponding IP address. The software architecture and validation measurements are presented together with possible future developments.

What can we learn from nested IoT low‐cost sensor networks for air quality? A case study of PM2.5 in Birmingham, UK

AbstractLow‐cost sensing and the Internet of Things (IoT), present new possibilities for unconventional monitoring of environmental parameters. This paper describes a series of intersecting networks of particulate matter sensors that were deployed across the Birmingham conurbation for a 12‐month period. The networks consisted of a combination of commercially available sensors and University developed sensors. Data from these networks were assimilated with data from a third‐party Zephyr deployment, along with the DEFRA AURN network, which was hosted on an open‐source online platform. This nesting of sensor networks allowed for new insights into sensor performance, including the accuracy of a large network to detect regional concentrations and the number of sensors needed for effective monitoring beyond indicative measurements. After comprehensive data validation steps, the sensors were shown to perform well during co‐location with reference instrumentation (exhibiting slopes of 0.74–1.3). The sensors demonstrated good capability of detecting temporal patterns of regional PM2.5 with the mean of the entire sensor network recording an annual mean PM2.5 concentration within 0.2 μgm−3 of the regulatory network annual mean observation. Network‐derived statistics for estimating urban background concentrations compared to a reference site increase in‐line with the number of sensors available, however when assessing this for near‐source concentrations the importance of sensor location rather than the number of sensors is highlighted. Overall, the network provided novel insights into local concentrations, detecting similar hotspots to those identified by a high‐resolution model. The increased spatial coverage afforded by the sensor network has the potential to support higher resolution evaluation of models and provide unprecedented spatial evidence for air pollution management interventions.

Smart environmental drone utilization for monitoring urban air quality

Urban air quality has significant and far-reaching impacts on both human health and the broader environment. Pollutants like particulate matter (PM2.5 and PM10), for instance, are associated with a range of health issues including respiratory conditions, asthma, heart diseases, and even contribute to low birth weight in newborns. These health implications extend to larger environmental concerns such as air pollution, greenhouse gas emissions, and global climate change. Recognizing the urgent need for effective and dynamic air quality monitoring solutions, this paper explores the use of smart environmental drones as a promising approach. Our drone is equipped with a state-of-the-art, low-cost particulate matter sensor that can accurately measure PM2.5 and PM10 concentrations. Operating at a flight speed of 10 m/s and capable of covering a range of 5 km, the drone executes a pre-programmed flight plan to autonomously map pollution levels across urban areas. With a 95% accuracy rate in sensor readings, our model significantly minimizes potential errors commonly associated with traditional air quality monitoring methods. Furthermore, it simplifies maintenance procedures, reducing both time and financial costs. By employing drone technology in this innovative manner, our model offers a cost-effective, reliable, and dynamic solution for monitoring urban air quality. It provides real-time, actionable pollution indices that can inform public health decisions, regulatory policies, and community awareness, thereby contributing to the broader goal of improving air quality and public health.

Long-term evaluation of commercial air quality sensors: an overview from the QUANT (Quantification of Utility of Atmospheric Network Technologies) study

Abstract. In times of growing concern about the impacts of air pollution across the globe, lower-cost sensor technology is giving the first steps in helping to enhance our understanding and ability to manage air quality issues, particularly in regions without established monitoring networks. While the benefits of greater spatial coverage and real-time measurements that these systems offer are evident, challenges still need to be addressed regarding sensor reliability and data quality. Given the limitations imposed by intellectual property, commercial implementations are often “black boxes”, which represents an extra challenge as it limits end users' understanding of the data production process. In this paper we present an overview of the QUANT (Quantification of Utility of Atmospheric Network Technologies) study, a comprehensive 3-year assessment across a range of urban environments in the United Kingdom, evaluating 43 sensor devices, including 119 gas sensors and 118 particulate matter (PM) sensors, from multiple companies. QUANT stands out as one of the most comprehensive studies of commercial air quality sensor systems carried out to date, encompassing a wide variety of companies in a single evaluation and including two generations of sensor technologies. Integrated into an extensive dataset open to the public, it was designed to provide a long-term evaluation of the precision, accuracy and stability of commercially available sensor systems. To attain a nuanced understanding of sensor performance, we have complemented commonly used single-value metrics (e.g. coefficient of determination, R2; root mean square error, RMSE; mean absolute error, MAE) with visual tools. These include regression plots, relative expanded uncertainty (REU) plots and target plots, enhancing our analysis beyond traditional metrics. This overview discusses the assessment methodology and key findings showcasing the significance of the study. While more comprehensive analyses are reserved for future detailed publications, the results shown here highlight the significant variation between systems, the incidence of corrections made by manufacturers, the effects of relocation to different environments and the long-term behaviour of the systems. Additionally, the importance of accounting for uncertainties associated with reference instruments in sensor evaluations is emphasised. Practical considerations in the application of these sensors in real-world scenarios are also discussed, and potential solutions to end-user data challenges are presented. Offering key information about the sensor systems' capabilities, the QUANT study will serve as a valuable resource for those seeking to implement commercial solutions as complementary tools to tackle air pollution.

Feasibility of low-cost particulate matter sensors for long-term environmental monitoring: Field evaluation and calibration

Low-cost sensor networks offer the potential to reduce monitoring costs while providing high-resolution spatiotemporal data on pollutant levels. However, these sensors come with limitations, and many aspects of their field performance remain underexplored. During October to December 2023, this study deployed two identical low-cost sensor systems near an urban standard monitoring station to record PM2.5 and PM10 concentrations, along with environmental temperature and humidity. Our evaluation of the monitoring performance of these sensors revealed a broad data distribution with a systematic overestimation; this overestimation was more pronounced in PM10 readings. The sensors showed good consistency (R2 > 0.9, NRMSE<5 %), and normalization residuals were tracked to assess stability, which, despite occasional environmental influences, remained generally stable. A lateral comparison of four calibration models (MLR, SVR, RF, XGBoost) demonstrated superior performance of RF and XGBoost over others, particularly with RF showing enhanced effectiveness on the test set. SHAP analysis identified sensor readings as the most critical variable, underscoring their pivotal role in predictive modeling. Relative humidity consistently proved more significant than dew point and temperature, with higher RH levels typically having a positive impact on model outputs. The study indicates that, with appropriate calibration, sensors can supplement the sparse networks of regulatory-grade instruments, enabling dense neighborhood-scale monitoring and a better understanding of temporal air quality trends.

Multi-Scenario Validation and Assessment of a Particulate Matter Sensor Monitor Optimized by Machine Learning Methods.

The aim was to evaluate and optimize the performance of sensor monitors in measuring PM2.5 and PM10 under typical emission scenarios both indoors and outdoors. Parallel measurements and comparisons of PM2.5 and PM10 were carried out between sensor monitors and standard instruments in typical indoor (2 months) and outdoor environments (1 year) in Shanghai, respectively. The optimized validation model was determined by comparing six machining learning models, adjusting for meteorological and related factors. The intra- and inter-device variation, measurement accuracy, and stability of sensor monitors were calculated and compared before and after validation. Indoor particles were measured in a range of 0.8-370.7 μg/m3 and 1.9-465.2 μg/m3 for PM2.5 and PM10, respectively, while the outdoor ones were in the ranges of 1.0-211.0 μg/m3 and 0.0-493.0 μg/m3, correspondingly. Compared to machine learning models including multivariate linear model (ML), K-nearest neighbor model (KNN), support vector machine model (SVM), decision tree model (DT), and neural network model (MLP), the random forest (RF) model showed the best validation after adjusting for temperature, relative humidity (RH), PM2.5/PM10 ratios, and measurement time lengths (months) for both PM2.5 and PM10, in indoor (R2: 0.97 and 0.91, root-mean-square error (RMSE) of 1.91 μg/m3 and 4.56 μg/m3, respectively) and outdoor environments (R2: 0.90 and 0.80, RMSE of 5.61 μg/m3 and 17.54 μg/m3, respectively), respectively. Sensor monitors could provide reliable measurements of PM2.5 and PM10 with high accuracy and acceptable inter and intra-device consistency under typical indoor and outdoor scenarios after validation by RF model. Adjusting for both climate factors and the ratio of PM2.5/PM10 could improve the validation performance.