COMPARATIVE ANALYSIS OF METHODS AND TOOLS FOR AMBIENT AIR QUALITY MONITORING
This paper presents contemporary approaches to ambient air quality monitoring in the vicinity of energy facilities, with a main focus on the use of intelligent systems and cutting-edge technologies. The primary objective of the study is to identify promising solutions for the development of intellig...
Збережено в:
| Дата: | 2025 |
|---|---|
| Автори: | , |
| Формат: | Стаття |
| Мова: | Англійська |
| Опубліковано: |
General Energy Institute of the National Academy of Sciences of Ukraine
2025
|
| Теми: | |
| Онлайн доступ: | https://systemre.org/index.php/journal/article/view/908 |
| Теги: |
Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
|
| Назва журналу: | System Research in Energy |
| Завантажити файл: |  |
Репозитарії
System Research in Energy| _version_ | 1871104418379726848 |
|---|---|
| author | Kulykivskyi, Oleksandr Ponomarenko, Oleksandr |
| author_facet | Kulykivskyi, Oleksandr Ponomarenko, Oleksandr |
| author_institution_txt_mv | [
{
"author": "Oleksandr Kulykivskyi",
"institution": null
},
{
"author": "Oleksandr Ponomarenko",
"institution": null
}
] |
| author_sort | Kulykivskyi, Oleksandr |
| baseUrl_str | https://systemre.org/index.php/journal/oai |
| collection | OJS |
| datestamp_date | 2026-07-18T12:57:49Z |
| description | This paper presents contemporary approaches to ambient air quality monitoring in the vicinity of energy facilities, with a main focus on the use of intelligent systems and cutting-edge technologies. The primary objective of the study is to identify promising solutions for the development of intelligent information and measurement systems capable of tracking air parameter variations in real time, analyzing pollution levels, and supporting the decision-making process. This paper emphasizes key criteria for assessing air pollution, including the concentration of harmful substances (PM2.5, PM10, NO2, SO2, CO, O3), the air quality index, toxic emissions, and natural meteorological factors. Modern methods for data collection and processing are examined, including Internet of Things IoT- based systems, cloud platforms, optical image analysis methods, and artificial intelligence such as machine learning and deep neural networks. Special attention is given to the application of regression models and hybrid approaches (CNN+LSTM) for PM2.5 level forecasting, enabling high-accuracy estimations based on both meteorological data and visual inputs. The study also describes stationary monitoring systems, their architecture, operational principles, and implementation examples using LoRa, LPWA, sensor networks, and mobile platforms. The results demonstrate the high efficiency of integrating artificial intelligence, big data, and the Internet of Things into monitoring systems, revealing new opportunities for the modernization of air quality monitoring and environmental risk management in areas with significant anthropogenic impact. |
| doi_str_mv | 10.15407/srenergy2025.03.081 |
| first_indexed | 2026-03-24T02:03:36Z |
| format | Article |
| fulltext |
Системні дослідження в енергетиці. 2025. 3(83) 81
ISSN 2786-7102 (Online), ISSN 2786-7633 (Print)
https://doi.org/10.15407/srenergy2025.03.081
UDK 620.179.14
Oleksandr Kulykivskyi*, https://orcid.org/0009-0006-8593-8795
Oleksandr Ponomarenko, PhD (Engin.), Associate Professor, https://orcid.org/0000-0002-
6538-0468
General Energy Institute of NAS of Ukraine, 172, Antonovycha St., Kyiv, 03150, Ukraine
*Corresponding author: sashkovskiy99@gmail.com
_____________________________________________________________________________________
COMPARATIVE ANALYSIS OF METHODS AND TOOLS FOR AMBIENT
AIR QUALITY MONITORING
Abstract. This paper presents contemporary approaches to ambient air quality monitoring in the vicinity
of energy facilities, with a main focus on the use of intelligent systems and cutting-edge technologies. The
primary objective of the study is to identify promising solutions for the development of intelligent
information and measurement systems capable of tracking air parameter variations in real time, analyzing
pollution levels, and supporting the decision-making process. This paper emphasizes key criteria for
assessing air pollution, including the concentration of harmful substances (PM2.5, PM10, NO2, SO2, CO,
O3), the air quality index, toxic emissions, and natural meteorological factors. Modern methods for data
collection and processing are examined, including Internet of Things IoT- based systems, cloud platforms,
optical image analysis methods, and artificial intelligence such as machine learning and deep neural
networks. Special attention is given to the application of regression models and hybrid approaches
(CNN+LSTM) for PM2.5 level forecasting, enabling high-accuracy estimations based on both
meteorological data and visual inputs. The study also describes stationary monitoring systems, their
architecture, operational principles, and implementation examples using LoRa, LPWA, sensor networks,
and mobile platforms. The results demonstrate the high efficiency of integrating artificial intelligence, big
data, and the Internet of Things into monitoring systems, revealing new opportunities for the modernization
of air quality monitoring and environmental risk management in areas with significant anthropogenic
impact.
Keywords: monitoring, intelligent systems, sensor networks, cloud technologies, machine learning,
forecasting.
1. Introduction
Air pollution remains one of the most pressing environmental challenges of our time, with significant
consequences for both human health and the natural ecosystem. Since air quality is closely linked to economic
activity, industrial emissions, transportation networks, and range of other human-induced factors, the ability
to effectively monitor its condition is becoming an essential part of modern environmental management
strategies. Today, a wide variety of methods are available for measuring and controlling air polution; however,
none of them can be considered truly universal or entirely effective. This creates certain challenges when it
comes to responding promptly to changes in ambient air quality.
One of the major limitations of existing air quality monitoring systems lies in their high costs-both in
terms of equipment and ongoing maintenance. Additionally, the coverage area of stationary monitoring
stations is often limited, making it difficult to obtain a comprehensive view of environmental conditions across
larger regions. Real-time data collection and processing also pose significant challenges, which, in turn,
complicate timely decision-making regarding the implementation of pollution reduction measures. Moreover,
many current forecasting methods fall short in delivering the necessary accuracy, largely due to the limitations
of traditional approaches in handling and interpreting large-scale environmental datasets [1].
In recent years, usage of machine learning models and neural networks in the fields of environmental
monitoring has emerged as a promising to addressing many of the existing challenges. These technologies can
process vast volumes of data, automate both measurement and forecasting tasks, and even optimize monitoring
routes with the help of unmanned aerial vehicles (UAVs) and other autonomous systems. Integrating such
methods not only enhances the efficiency of monitoring efforts but also significantly improves the quality of
decision-making by providing accurate and timely information about the state of ambient air.
https://orcid.org/0009-0006-8593-8795
mailto:sashkovskiy99@gmail.com
Системні дослідження в енергетиці. 2025. 3(83) 82
In this context, the advancement and implementation of the new technologies in air pollution monitoring
systems-particularly those involving machine learning-hold great potential to transform current practices.
These innovations enable faster responces to environmental changes and contribute to more effective resource
management aimed at protecting public health and preserving the natural environmental [2].
The objective of the study is to identify the most promising solutions for the development of intelligent
information and measurement systems design to monitor air quality near energy facilities. These systems
should be capable of recognizing patterns in air quality changes in real time and support the decision-making
process in environmental management.
2. Analysis of methods for diagnosing ambient air quality
Various types of control strategies and a wide range of analytical methods can be employed to assess
the quality of the ambient air. In recent years, the rapid development of artificial intelligence has introduced
new and increasingly sophisticated forecasting techniques. The key criteria for determining air pollution levels
include:
• Pollutant concentration, such as: particular matter (PM2.5 and PM10), nitrogen dioxide, sulfur dioxide,
ozone, carbon dioxide, carbon monoxide and volatile organic compounds.
• Air Quality Index (AQI). This integrated indicator aggregates data on the concentration of major
pollutants, offering an overall assessment of air quality. It`s typically calculated based on the levels of PM2.5,
NO2, SO2, O3, CO and provides a rating scale ranging from “good” to “very poor”.
• Presence of toxic emissions. Here an assessment is made of emissions that may contain toxic
components, such as carcinogens, heavy metals etc.
• Natural environment factors. The temperature, atmospheric pressure, humidity, which can influence
pollution dispersion and concentration levels.
The current methods for measuring and analyzing air quality can be classified into several categories
(Fig. 1).
Fig. 1. Classification of methods for measuring and analyzing ambient air quality
In modern monitoring of an ambient air quality, laboratory-based methods continue to play a crucial
role due to their high precision and reliability in data collecting, however, they are less flexible and slower in
real-time supporting and decision-making processes. Laboratory-based approaches rely on well-established
traditional techniques, which include optical, acoustic, chemical, thermal and electrical methods. Optical
methods encompass technologies like thermographic imaging, infrared spectroscopy, laser spectroscopy, laser
scanning, photometry and etc. Acoustic methods primarily involve acoustic emission analysis. Chemical
Системні дослідження в енергетиці. 2025. 3(83) 83
methods are represented by gas chromatography, while thermal approaches involve the use of thermal cameras,
particularly thermographic systems, which are used to detect thermal anomalies that may indicate elevated
concentrations of pollutants [3].
Among the current and reviewed solutions there`re also a predictive methods for measuring ambient air
quality, these methods are based on models and systems capable of operating large data on pollutant
concentration and local meteorological conditions. This includes mathematical modeling, machine learning
models, real-time monitoring systems, which integrate data from multiple sources to predict the spread of the
iar pollution. For example, Internet of Things (IoT) systems, mobile applications, AI-powered platforms that
operate Big Data analytics are being used for now to deliver highly accurate air quality forecasts [4, 5].
Let`s deeply examine the current system`s features of the different monitoring and forecasting systems
for air quality.
IoT-based systems for monitoring ambient air quality operate through distributed sensor networks,
which continuously collects pollution data with further transmitting it to the analytical processing systems.
These systems include both stationary and mobile measuring gadgets, which are equipped with specialized
sensors for detecting concentrations of PM2.5 and PM10, harmful gases such as CO₂, NO₂, SO₂, and O₃, as
well as key meteorological indicators. Data transmition is carried by wireless communication technologies
(Wi-Fi, GSM, LoRa) to cloud-based analytical platforms (AWS, ThingSpeak), where advanced machine
learning and mathematical modeling are being applied in order to provide a comprehensive assessment of air
quality conditions. Also, the integration of AI technologies, distributed ledger systems, and Big Data
processing significantly enhances the accuracy of analytical results and ensures the transparency and
objectivity of the monitoring process [6−8].
Modern IoT solutions for monitoring ambient air quality, such as AIRO and Make Em` Green, are
known for their energy efficiency and low cost, and also with their capability of providing real-time data,
which contributes to the improvement of the environmental monitoring system and expands the possibilities
of supporting management decision-making [6, 7]. Fig. 2 illustrates the architecture of the AIRO air quality
monitoring system.
Fig. 2. System architecture of the AIRO air quality monitoring solution based on multiple device prototypes deployed
across various GPS locations [6]
The Real-Time Air Quality Monitoring Model using Fuzzy Inference System combines IoT
technologies and fuzzy logic to enable continuous data aqusition of the air quality [9, 10]. It automatically
collects all the data about all major pollutants, such as CO, NO₂, PM₁₀, SO₂, O3, then analyzes it by using a set
of fuzzy rules along with centroid type defuzzification, which helps to detect and classify the air quality into
categories such as “good”, “moderate”, “poor”. Thanks to IoT sensors, the system offers quick responsiveness,
Системні дослідження в енергетиці. 2025. 3(83) 84
operational flexibility and capacity to forecast changes in environmental conditions, which makes it
particularly well suited for usage in dynamic urban settings.
3. Optical methods for assessing air pollution based on image analysis
Air pollution causes significant effect on the atmosphere and often leads to reduced visibility in
different areas. In particular, dust particles create a visible haze, which may be observed with the naked eye.
That`s why the usage of optical methods for assessing air quality has gained considerable attention in nowadays
scientific research. Studies in this field of science focus on the development of the novel image processing
algorithms and advancement of deep learning models for assessing the AQI.
Visual based approaches for estimation of PM2.5 concentration. Wang and co-authors in their
research [11] proposed a method of using visual information to estimate PM2.5 concentrations by exploring
the statistical correlation between light absorption and pollution levels. Their approach is based on
understanding that atmospheric aerosols and gaseous particles absorb light and cause reduced visibility. The
researchers applied Lambert-Beer`s light absorption model, which represents the relationship among three
variables at each pixel: the distance between the observer and the object and the light absorption coefficient.
In their algorithm Dark Channel Prior model has been used, which is based on observation that in a haze-free
image, at least one colour channel contains pixels with very low intensities close to zero.
Hierarchical algorithms and machine learning approaches. Yao and co-authors [12] developed
hierarchical structural algorithm, which integrates low-level image features with the higher-level optical
features. Their method uses the sky colour shift (the difference between good and bad weather in sky) and
illumination gradients, which reflects the extent of solar light scattering in the atmosphere. Unlike the other
systems that analyze only the images from the fixed locations, Yao`s algorithm is capable of selecting optimal
combinations of atmospheric indices, even in complex urban scenes.
To estimate PM2.5 researchers used Support Vector Regression (SVR). Their dataset comprises 2945
daytime photographs captured by smartphones with varying resolutions and scenes (lakes, mountains, urban
areas). Photos taken during rain and snow were excluded. Every photo was associated with a PM2.5 value
measured by the Nature Clean sensor, which operates on the principle of laser light scattering. Figure 3
demonstrates photos, which were collected via crowdsourcing platform “Moji Weather” − weather-focused
social media platform where users post geotagged photos. For gathering these photos, published in Beijing, a
web crawling technique was used.
Fig. 3. Spatial distribution of photos collected through crowdsourcing from the Moji Weather app (a), and
examples of photos taken in three districts (b) [12]
Системні дослідження в енергетиці. 2025. 3(83) 85
In a similar approach, Liu Chenbin and colleagues [13] also developed a support vector regression model
to predict PM2.5 levels using six specific image features: contrast, entropy, dark channel, sun position, sky
gradient and blue colour component. For optimization the selected features, the researchers employed
techniques such as Principal Component Analysis (PCA) and sequential feature selection. Their study was
based on datasets collected in Beijing and Shanghai (China), and Phoenix (USA), comprising a total of 327,
1954 and 4306 daytime images. These images were taken from fixed locations and corresponding weather
conditions and PM2.5 index data were obtained from the nearest meteorological stations.
Proposed methods demonstrate the potential of leveraging optical data and image analysis for air quality
assessment, which are offering a valuable complement to traditional atmospheric pollution monitoring
methods.
4. Application of machine learning and neural networks in air quality forecasting
Artificial intelligence (AI) and it`s subfields, such as machine learning and deep neural networks, play
a crucial role in addressing complex environmental challenges. One of the key areas of implementation for
these technologies is air quality forecasting, which enables timely responses to changes in environmental
conditions and supports implementation of effective measures to reduce pollution levels. Due to their ability
to analyze large volumes of data and identify complex patterns, machine learning algorithms provide high
prediction accuracy and contribute to the development of environmental strategies in urban areas.
Application of machine learning in air quality prediction. One of the approaches for forecasting air level
pollution levels involves the use of machine learning algorithms within specialized monitoring systems. For
example, in [14], an interactive system for predicting urban air quality was proposed, based on machine
learning methods such as Random Forest, Support Vector Regressor, and CatBoost. This system uses real-
world environmental data, including PM2.5, PM10, and nitrogen oxide levels, to generate accurate air quality
predictions and provide real-time updates. One of the main advantages of this model is its ability to integrate
data from different cities around the world, thus supporting global air pollution monitoring.
Another study [5] investigates the application of the LSTM-E model to predict PM2.5 concentrations.
The authors highlight that this model demonstrates high efficiency in capturing the spatiotemporal
dependencies in the data, which significantly improves the forecasting accuracy compared to traditional
approaches such as TDNN, STDL, SVR and ARMA. The study showed that the inclusion of auxiliary
meteorological data improves the forecast quality, and deep neural networks, in particular LSTM, significantly
outperform conventional methods in forecasting accuracy.
Another interesting model is APNet, which uses convolutional neural networks (CNN) together with
long-short-term memory (LSTM) networks [15]. This model was created to predict PM2.5 levels in Beijing.
It uses data on wind speed, rainfall, and pollution levels over the past 24 hours. Tests have shown that APNet
performs better than other popular machine learning methods, such as SVM, random decision forests (RD),
decision trees (DT), multilayer perceptron (MLP), simply CNN, or simply LSTM. It produced the most
accurate results with the lowest prediction errors. The authors say that using CNN to look at spatial data and
LSTM to learn temporal patterns is a good way to make air pollution forecasts more accurate.
In the study [16], researchers used different machine learning methods like Random Forest, XGBoost,
and deep neural networks to predict PM2.5 levels in Tehran. A very important part of the study was using
remote sensing data (AOD) together with weather data, which helped to improve prediction accuracy. One big
problem was the lack of data, especially AOD03, which made modeling harder. The authors said that XGBoost
gave the best results because it is fast and can find the most important environmental factors.
Finally, the review paper [17] looked at how different machine learning methods are used to study air
pollution. It included methods like BPNN, XGBoost, Random Forest (RF), Continuous Wavelet Transform
(CWT), Decision Trees (DT), k-Nearest Neighbors (k-NN), and hybrid models. The authors said that using a
mix of methods can give more accurate results and help make predictions faster. The review also talked about
the future use of federated learning, transfer learning, and automatic parameter tuning using genetic algorithms.
Системні дослідження в енергетиці. 2025. 3(83) 86
It also said that we need new models to predict air pollution during natural disasters, like fires, and military
conflicts, which is very important for our country today.
5. Stationary air quality monitoring systems
Stationary air quality monitoring systems play a pivotal role in detecting and analyzing environmental
air pollution levels. Thanks to usage of different sensors and technologies of data transmission, these systems
allow assessing pollution level in real time and allow to make special decisions for improvement the current
environmental situation. In recent years the development of wireless technologies and optimization algorithms
has made these systems more effective.
One of the many approaches of realization wireless technologies in stationary-based measurement
systems is implementation it in the city streetlights network. This system has sensors that measure the amount
of particles in the the air using light scattering. Wireless data transfer makes it easy to scale the system and
connect it with city infrastructure. This allows for fast and continuous data collection and helps with ecological
analysis and decision-making [18].
Another promising approach involves the implementation of Low Power Wide Area (LPWA)
technology, which is used for air quality monitoring across large geographical areas. The sensors in such
systems are capable of measuring key environmental parameters, including CO₂ concentration, PM2.5 levels,
temperature, and humidity. The collected data is transmitted via LoRaWAN to centralized servers, where it is
processed and visualized for further analysis. This enables efficient tracking of pollution level fluctuations
even in remote or hard-to-reach areas, while maintaining minimal energy consumption, making the system
suitable for long-term deployment [19]. Figure 4 presents the architecture of an LPWA-based air quality
monitoring system.
Fig. 4. Architecture of an LPWA-based air quality monitoring system [19]
Another promising system involves the use of LoRa technology for air quality monitoring within urban
areas, particularly in university campuses [20]. These systems include the deployment of sensors at multiple
locations across the city, with data transmitted to cloud platforms for further analysis. This technology can be
Системні дослідження в енергетиці. 2025. 3(83) 87
further enhanced through integration with mobile platforms such as drones or vehicles, which increases data
collection accuracy, enables monitoring in hard-to-reach areas, and supports rapid response to changes in air
quality.
Sensor placement optimization methods. Optimizing the placement of sensors is a key factor in
improving the effectiveness of monitoring systems. One approach involves using a genetic algorithm to
strategically distribute sensors across microregions with similar air pollution characteristics. This method helps
to minimize data variation and increases the accuracy of pollution estimates. Spatial interpolation, used in
conjunction with this approach, allows for the creation of a complete pollution map of the study area, thus
supporting more informed decision-making [21].
Another method to improve the performance of stationary monitoring systems involves optimizing the
placement of sensors based on information theory. Some studies use the Lagrangian atmospheric dispersion
model to estimate pollution levels under different meteorological conditions. This allows determining the
optimal locations for sensor placement, ensuring maximum coverage of the territory and high accuracy of the
collected data. This approach significantly improves the quality of ambient air monitoring and increases the
effectiveness of air pollution control measures [22].
6. Dynamic air quality monitoring systems
Due to increasing air pollution levels, in the current context, dynamic monitoring systems play a crucial
role by enabling measurements while in motion, covering large areas and providing real-time data. Such
systems include unmanned aerial vehicles (UAVs) and sensor modules installed on public transportation.
These technologies significantly expand the capabilities of environmental monitoring by detecting pollution
sources and analyzing their dynamics in real time.
Air pollution monitoring using mobile sensors on the vehicles. Another promising method of dynamic
monitoring involves using sensors placed on public transport vehicles, such as city buses [23]. This system
allows data collection on pollution levels in various parts of the city without the need to install a large number
of stationary stations.
For example mobile sensors placed on buses measure concentrations of PM2.5, PM10, CO₂, and other
pollutants while the vehicle is in motion. The data are transmitted in real time to servers for further analysis,
enabling the creation of interactive pollution maps and tracking changes in air quality. Experimental testing of
this technology in urban environments has shown that the use of mobile sensor platforms significantly
improves the accuracy of pollution mapping. This makes it possible to assess the impact of transport, industrial
facilities, and other emission sources on air quality. In addition, this method is more cost-effective compared
to the deployment of an extensive network of fixed monitoring stations [23].
Use of UAVs for air quality monitoing. UAVs are increasingly used in monitoring due to their ability to
operate in hard-to-reach areas and perform three-dimensional pollution mapping [24]. Studies show that using
UAVs for monitoring significantly increases measurement accuracy compared to stationary stations, especially
in cases of local pollution sources [25].
One example is a monitoring system that includes particulate sensors, temperature and humidity sensors,
and barometric pressure sensors mounted on UAVs. This system allows measurements of air quality
parameters at various altitudes (from 0 to 120 meters), contributing to a more detailed analysis of the vertical
distribution of pollutants. Research shows that integrating UAVs with IoT platforms enables the acquisition
of accurate environmental data in real time and their transmission for further processing.
Moreover, UAVs are effective in detecting pollution patterns and forecasting changes in air quality,
which is particularly important for urban areas. However, the accuracy of collected data largely depends on
sensor characteristics, their placement, and payload weight. Therefore, it is important to integrate UAV-
collected data with other sources to improve forecasting precision [26, 27].
Table 1 summarizes the advantages and disadvantages of various monitoring systems.
Системні дослідження в енергетиці. 2025. 3(83) 88
Table 1. Advantages and disadvantages of air quality monitoring systems
System Informative parameter Advantages Disadvantages
AIRO [6] Air quality index Mobility, decentralization,
accessibility, flexibility in
implementation
Power consumption, Limited
accuracy, dependence on cloud
infrastructure
Monitoring system
based on wireless
sensor networks [9]
PM2,5 i PM10 Scalability, High reliability Implementation cost, dependence on
data transmission technologies,
constant sensor calibration, limited
sensor accuracy
Make Em` Green
[7]
PM2,5 i PM10
СО2, temperature,
humidity
Real-time data acquisition, budget-
friendly, energy-saving, easy to
configure
Limited accuracy compared to
professional fixed stations,
dependence on stable Wi-Fi
connection for data transfer
Real-Time Air
Quality Monitoring
Model using Fuzzy
Inference System
[10]
pollutant concentrations,
CO, NO2, PM10, SO2
Real-time monitoring
Effective pollution management in
urban areas
Forecasting using fuzzy logic
Complexity of setting up and
configuring a fuzzy system
Requires stable internet connection
for IoT to work
Kumar Sai`s
monitoring system
[14]
Ammonia, Carbon dioxide,
smoke, carbon monoxide
Cheapness, accessibility, use of IoT
for environmental monitoring
Vulnerable to external factors such
as temperature and humidity
Gresha Batia – air
quality monitoring
system [4]
PM2,5 and PM10
Nitrogen oxides (NOx)
Accurate prediction of air quality
pollution,
Timely response to changing
environmental conditions
Integration with Air Quality
Programmatic API
The system depends on the quality
of sensor data, errors
The need for more powerful
computing resources to process
large data sets
LPWA-Based Air
Quality Monitoring
System [19]
pollutant concentrations,
PM2.5, CO2.
Energy Efficiency
Wide Coverage
Scalability
Data transfer limitations
Dependence on LPWA
infrastructure
Need for periodic sensor calibration
Real-time air
pollution
monitoring with
sensors on city bus
[23]
pollutant concentrations,
PM2.5, PM10, CO2.
Cost Reduction
Real-time Monitoring
Mobility and Reach
Accuracy
Limitations in pollutant types
Use of LoRa [20] pollutant concentrations,
PM2.5, CO2
Temperature and humidity
Energy efficiency
Scalability
Data transmission range
Minimal infrastructure
Range limitations in urban
environments
Lack of mobility
Data transmission delay
Limitations in the number of sensors
Real-time vertical
air quality
monitoring system
pollutant concentrations,
PM2.5, PM10
Temperature and humidity
Pressure (barometric
sensor)
High accuracy and detail
Integration with IoT
Flexibility
UAV flight time limitations
Weather dependence
Data transmission and storage issues
Monitoring ambient
air pollution using
UAVs
pollutant concentrations
PM2.5, PM10, NO2, CO2
Mobility and accessibility
Speed of data collection
Flexibility and scalability
Data Limits
Autonomy
The results of the analysis of the advantages and disadvantages of the proposed methods for improving
monitoring are systematized in Table 2.
Системні дослідження в енергетиці. 2025. 3(83) 89
Table 2. Advantages and disadvantages of the proposed methods for monitoring ambient air quality
Method Informative parametr Advantages Disadvantages
AirVision [11] PM2.5 concentration
Color channels for haze
level estimation
Non-contact method
Cameras provide real-time operation
Using deep learning to improve
accuracy
Dependence on weather conditions
Need for calibration for different
locations
Nature Clean [12] PM2,5 High accuracy
Adaptability
Using real-life images
Weather restrictions
Dependence on image quality
Need for a large dataset for more
accurate model training
LSTME model [5] PM2,5 High forecast accuracy
Multi-station forecasting
Need for large amounts of data to
train a forecasting model
Computing resource consumption
Regular model updates
Dependence on input parameters
APNet Neural
network based on
CNN-LSTM [15]
PM2,5 High accuracy, practicality
Improving the efficiency of pollution
management
Dependence on training data
Restrictions by geographic region
Modeling complexity
Image-Processing
Algorithms [12, 13]
PM2,5 Image optimization to improve
forecasts
Multi-station forecasting
Integrating data from different
sources
Lighting limitations
Limitations in the choice of shooting
locations
Technical complexity in
calculations
Optimizing sensor
placement [21]
Sensor positioning Optimize sensor placement
Reduce data distribution variation
Support management decisions
Resource saving
Dependence on the accuracy of the
initial data
Complexity in data collection and
processing
High requirements for computing
resources
Forecasting model
Apache Spark [17]
pollutant concentrations,
PM2.5, NO2, CO2.
High prediction accuracy
Processing speed
Scalability
Combining machine learning
methods
High computational requirements
Long model training time
Lack of universal approaches
7. Conclusions
Based on the conducted analysis, it can be concluded that currently there is no universal method for
comprehensive and high-quality monitoring of atmospheric air that would allow simultaneous detection of all
pollutants and negative impacts. Most air quality parameters are controlled through the deployment of
stationary observation stations or by collecting samples at various locations within a city or region. This
requires the use of a large amount of specialized equipment and is often accompanied by the need to carry out
complex data collection procedures, such as connecting to air station networks or traveling to hard-to-reach
areas. All of this complicates and delays the monitoring process, making it labor-intensive and costly.
Since most traditional atmospheric monitoring methods have certain technical difficulties in being
implemented in urban conditions with a large number of pollution sources, one of the most promising methods
is environmental air pollution monitoring using UAVs, as well as the use of machine learning or neural
networks to optimize the measurement process, determine UAV flight routes, and process the obtained data.
This not only reduces costs but also significantly increases the accuracy and efficiency of monitoring.
References
1. Zaporozhets, A., Babak, V., Popov, O., Scherbak, L., & Kuts, Yu. (2023). Methodological Support of Air Pollution
Monitoring System. In A. Zaporozhets (Ed.), Systems, Decision and Control in Energy V. Studies in Systems,
Decision and Control, 481. Springer, Cham. https://doi.org/10.1007/978-3-031-35088-7_41
https://doi.org/10.1007/978-3-031-35088-7_41
Системні дослідження в енергетиці. 2025. 3(83) 90
2. Vyshnevskyi, A.V., & Potapova, N.A. (2024, November 1). Methods of data analysis in an information system for
air pollution indicators forecasting. III International Scientific and Practical Conference "Applied Aspects of
Contemporary Interdisciplinary Research". Vasyl' Stus Donetsk National University, Vinnytsia, Ukraine. Retrieved
November 10, 2025, from https://jpasmd.donnu.edu.ua/article/view/16741 [in Ukrainian].
3. Posudin, Yu.I. (2012). Environmental monitoring with the basics of metrology [Textbook]. Кyiv, Ukraine (426 p.).
Retrieved November 10, 2025, from https://ekmair.ukma.edu.ua/items/ce974438-d0fc-458b-8c4a-cc69e16825e8
[in Ukrainian].
4. Bhatia, G. (2024). Predictive urban air quality monitoring for healthier cities. South Eastern European Journal of
Public Health (SEEJPH), 25(S1), 1627–1634. https://doi.org/10.70135/seejph.vi.2166
5. Li, X., Peng, L., Yao, X., Cui, S., You, C., & Chi, T. (2017). Long short-term memory neural network for air
pollutant concentration predictions: Method development and evaluation. Environmental Pollution, 231, 997–1004.
https://doi.org/10.1016/j.envpol.2017.08.114
6. Kumar, T., & Doss, A. (2023). AIRO: Development of an Intelligent IoT-based Air Quality Monitoring Solution for
Urban Areas. Procedia Computer Science, 218, 262−273. https://doi.org/10.1016/j.procs.2023.01.008
7. Nisanthan, G., & Shanaz, A. L. F. (2024). IoT-based Air Quality Monitoring and Prediction System. Journal of
Information and Communication Technology (JICT), 2(1), 1−6. Retrieved November 10, 2025, from
https://www.researchgate.net/publication/388103821_IoT-
based_Air_Quality_Monitoring_and_Prediction_System
8. Rahaman, T. (2025). Smart Environmental Monitoring Systems for Air and Water Quality Management. American
Journal of Advanced Technology and Engineering Solutions, 1(1), 1−22. https://doi.org/10.63125/a08zay47
9. Gowda, M. P., G Y, H., K N, J., Srushti, & Padma, R. (2021). Air quality monitoring system. International Journal
of Engineering Research & Technology (IJERT), 9(12), 69−71. Retrieved November 10, 2025, from
https://www.ijert.org
10. Saleem, M., Shingari, N., Farooq, M., Mago, B., & Khan, M. (2024). Real-time air quality monitoring model using
fuzzy inference system. International Journal of Advanced Computer Science and Applications, 15(6), 838–846.
https://doi.org/10.14569/IJACSA.2024.0150684
11. Wang, H., Yuan, X., Wang, X., Zhang, Y., & Dai, Q. (2014). Real-time air quality estimation based on color image
processing. 2014 IEEE Visual Communications and Image Processing Conference. IEEE, Piscataway, NJ (pp. 326–
329). https://doi.org/10.1109/VCIP.2014.7051572
12. Yao, S., Wang, F., & Huang, B. (2022). Measuring PM2.5 concentrations from a single smartphone photograph.
Remote Sensing, 14(11), 2572. https://doi.org/10.3390/rs14112572
13. Liu, C., Tsow, F., Zou, Y., & Tao, N. (2016). Particle pollution estimation based on image analysis. PLOS ONE, 11,
e0145955. https://doi.org/10.1371/journal.pone.0145955
14. Sai, K. B. K., Mukherjee, S., & Sultana, H. P. (2019). Low cost IoT based air quality monitoring setup using Arduino
and MQ series sensors with dataset analysis. Procedia Computer Science, 167, 376–385.
https://doi.org/10.1016/j.procs.2020.01.043
15. Huang, C.-J., & Kuo, P.-H. (2018). A deep CNN-LSTM model for particulate matter (PM2.5) forecasting in smart
cities. Sensors, 18(7), 2220. https://doi.org/10.3390/s18072220
16. Joharestani, M. Z., Cao, C., Ni, X., Bashir, B., & Talebiesfandarani, S. (2019). PM2.5 prediction based on random
forest, XGBoost, and deep learning using multisource remote sensing data. Atmosphere, 10(7), 373.
https://doi.org/10.3390/atmos10070373
17. Agbehadji, I. E., & Obagbuwa, I. C. (2024). Systematic review of machine learning and deep learning techniques
for spatiotemporal air quality prediction. Atmosphere, 15(11), 1352. https://doi.org/10.3390/atmos15111352
18. Wang, L. (2024). Design industrial 5.1 air quality monitoring system and develop smart city infrastructure.
Measurement: Sensors, 35, 101292. https://doi.org/10.1016/j.measen.2024.101292
19. Zheng, K., Zhao, S., Yang, Z., Xiong, X., & Xiang, W. (2016). Design and implementation of LPWA-based air
quality monitoring system. IEEE Access, 4, 3238−3245. https://doi.org/10.1109/ACCESS.2016.2582153
20. Fadhil, A. A., Abdullah, F. Y., & Yaseen, M. T. (2024). Design and implementation of air pollution monitoring
system in Mosul city. Instrumentation Mesure Métrologie, 23(1), 71–77. https://doi.org/10.18280/i2m.230107
21. Borges, M. A., Lopes, P. B., & da Silva, L. A. (2023). Network optimization of carbon monoxide sensor nodes in
the metropolitan region of São Paulo. Electronics, 12(22), 4647. https://doi.org/10.3390/electronics12224647
22. Mano, Z., Kendler, S., & Fishbain, B. (2022). Information theory solution approach to the air pollution sensor
location–allocation problem. Sensors, 22(10), 3808. https://doi.org/10.3390/s22103808
23. Kaivonen, S., & Ngai, E. C.-H. (2019). Real-time air pollution monitoring with sensors on city bus. Digital
Communications and Networks, 6(1), 42–51. https://doi.org/10.1016/j.dcan.2019.03.003
24. Zaporozhets, A., Kovtun, S., & Dekusha, O. (2019, September). System for monitoring the technical state of heating
networks based on UAVs. Conference on Computer Science and Information Technologies (pp. 935−950). Springer
Cham. https://doi.org/10.1007/978-3-030-33695-0_61
25. Zaporozhets, A. (2017). Analysis of means for monitoring air pollution in the environment. Science-based
technologies, 35(3), 242−252 [in Ukrainian]. https://doi.org/10.18372/2310-5461.35.11844
26. Shukri, S., Rahiman, M., & Sahwee, Z. (2024). Real-time vertical air quality monitoring system development for
urban scale. International Journal of Geoinformatics, 20(8), 102. https://doi.org/10.52939/ijg.v20i8.3463
https://jpasmd.donnu.edu.ua/article/view/16741
https://ekmair.ukma.edu.ua/items/ce974438-d0fc-458b-8c4a-cc69e16825e8
https://doi.org/10.1016/j.envpol.2017.08.114
http://www.ijert.org/
https://doi.org/10.14569/IJACSA.2024.0150684
https://doi.org/10.1371/journal.pone.0145955
https://doi.org/10.1016/j.procs.2020.01.043
https://doi.org/10.3390/s18072220
https://doi.org/10.3390/atmos10070373
https://doi.org/10.3390/atmos15111352
https://doi.org/10.1016/j.measen.2024.101292
https://doi.org/10.1109/ACCESS.2016.2582153
https://doi.org/10.18280/i2m.230107
https://doi.org/10.3390/electronics12224647
https://doi.org/10.3390/s22103808
https://doi.org/10.1016/j.dcan.2019.03.003
https://doi.org/10.18372/2310-5461.35.11844
https://doi.org/10.52939/ijg.v20i8.3463
Системні дослідження в енергетиці. 2025. 3(83) 91
27. Sarode, G., Kokate, P., Middey, A., & Sadishtap, S. (2023). Review on drone-assisted air-quality monitoring
systems. Drones and Autonomous Vehicles, 1. https://doi.org/10.35534/dav.2023.10005
ПОРІВНЯЛЬНИЙ АНАЛІЗ МЕТОДІВ ТА ЗАСОБІВ
МОНІТОРИНГУ СТАНУ АТМОСФЕРНОГО ПОВІТРЯ
Олександр Куликівський*, https://orcid.org/0009-0006-8593-8795
Олександр Пономаренко, канд. техн. наук, доцент, https://orcid.org/0000-0002-6538-0468
Інститут загальної енергетики НАН України, вул. Антоновича, 172, Kиїв, 03150, Україна
*Автор-кореспондент: sashkovskiy99@gmail.com
Анотація. У роботі розглянуто сучасні підходи до моніторингу якості атмосферного повітря в
околі об’єктів енергетики з акцентом на використання інтелектуальних систем та новітніх
технологій. Основною метою дослідження є виявлення перспективних рішень для побудови
інтелектуальних інформаційно-вимірювальних систем, що дозволяють в реальному часі
відстежувати зміни параметрів повітря, аналізувати рівень забруднення та підтримувати
прийняття управлінських рішень. У роботі акцентовано увагу на основних критеріях оцінки
забрудненості повітря, таких як концентрація шкідливих речовин (PM2.5, PM10, NO₂, SO₂, CO, O₃),
індекс якості повітря, токсичні викиди та природні метеорологічні чинники. Розглянуто сучасні
методи збору та опрацювання даних, зокрема системи на базі Інтернету речей, хмарні
платформи, оптичні методи аналізу зображень, а також алгоритми штучного інтелекту,
включно з машинним навчанням та глибокими нейронними мережами. Особливу увагу приділено
використанню моделей регресії та гібридних підходів (CNN+LSTM) для прогнозування рівня PM2.5,
що дозволяє досягати високої точності оцінювання на основі як метеоданих, так і візуальної
інформації. Також описано стаціонарні системи моніторингу, їх архітектуру, принципи
функціонування та приклади реалізації з використанням LoRa, LPWA, сенсорних мереж і мобільних
платформ. Отримані результати свідчать про високу ефективність інтеграції штучного
інтелекту, великих даних та Інтернету речей у системах моніторингу, що відкриває нові
можливості для модернізації систем моніторингу якості повітря та управління екологічними
ризиками у зонах підвищеного техногенного навантаження.
Ключові слова: моніторинг, інтелектуальні системи, сенсорні мережі, хмарні технології, машинне
навчання, прогнозування.
Надійшла до редколегії: 03.06.2025
https://doi.org/10.35534/dav.2023.10005
https://orcid.org/0009-0006-8593-8795
mailto:sashkovskiy99@gmail.com
|
| id | systemreorg-article-908 |
| institution | System Research in Energy |
| keywords_txt_mv | keywords |
| language | English |
| last_indexed | 2026-07-19T01:23:47Z |
| publishDate | 2025 |
| publisher | General Energy Institute of the National Academy of Sciences of Ukraine |
| record_format | ojs |
| resource_txt_mv | systemreorg/86/26dfa31eec310697c8318535fabf4186.pdf |
| spelling | systemreorg-article-9082026-07-18T12:57:49Z COMPARATIVE ANALYSIS OF METHODS AND TOOLS FOR AMBIENT AIR QUALITY MONITORING Порівняльний аналіз методів та засобів моніторингу стану атмосферного повітря Kulykivskyi, Oleksandr Ponomarenko, Oleksandr monitoring, intelligent systems, sensor networks, cloud technologies, machine learning, forecasting. моніторинг, інтелектуальні системи, сенсорні мережі, хмарні технології, машинне навчання, прогнозування. This paper presents contemporary approaches to ambient air quality monitoring in the vicinity of energy facilities, with a main focus on the use of intelligent systems and cutting-edge technologies. The primary objective of the study is to identify promising solutions for the development of intelligent information and measurement systems capable of tracking air parameter variations in real time, analyzing pollution levels, and supporting the decision-making process. This paper emphasizes key criteria for assessing air pollution, including the concentration of harmful substances (PM2.5, PM10, NO2, SO2, CO, O3), the air quality index, toxic emissions, and natural meteorological factors. Modern methods for data collection and processing are examined, including Internet of Things IoT- based systems, cloud platforms, optical image analysis methods, and artificial intelligence such as machine learning and deep neural networks. Special attention is given to the application of regression models and hybrid approaches (CNN+LSTM) for PM2.5 level forecasting, enabling high-accuracy estimations based on both meteorological data and visual inputs. The study also describes stationary monitoring systems, their architecture, operational principles, and implementation examples using LoRa, LPWA, sensor networks, and mobile platforms. The results demonstrate the high efficiency of integrating artificial intelligence, big data, and the Internet of Things into monitoring systems, revealing new opportunities for the modernization of air quality monitoring and environmental risk management in areas with significant anthropogenic impact. У роботі розглянуто сучасні підходи до моніторингу якості атмосферного повітря в околі об’єктів енергетики з акцентом на використання інтелектуальних систем та новітніх технологій. Основною метою дослідження є виявлення перспективних рішень для побудови інтелектуальних інформаційно-вимірювальних систем, що дозволяють в реальному часі відстежувати зміни параметрів повітря, аналізувати рівень забруднення та підтримувати прийняття управлінських рішень. У роботі акцентовано увагу на основних критеріях оцінки забрудненості повітря, таких як концентрація шкідливих речовин (PM2.5, PM10, NO₂, SO₂, CO, O₃), індекс якості повітря, токсичні викиди та природні метеорологічні чинники. Розглянуто сучасні методи збору та опрацювання даних, зокрема системи на базі Інтернету речей, хмарні платформи, оптичні методи аналізу зображень, а також алгоритми штучного інтелекту, включно з машинним навчанням та глибокими нейронними мережами. Особливу увагу приділено використанню моделей регресії та гібридних підходів (CNN+LSTM) для прогнозування рівня PM2.5, що дозволяє досягати високої точності оцінювання на основі як метеоданих, так і візуальної інформації. Також описано стаціонарні системи моніторингу, їх архітектуру, принципи функціонування та приклади реалізації з використанням LoRa, LPWA, сенсорних мереж і мобільних платформ. Отримані результати свідчать про високу ефективність інтеграції штучного інтелекту, великих даних та Інтернету речей у системах моніторингу, що відкриває нові можливості для модернізації систем моніторингу якості повітря та управління екологічними ризиками у зонах підвищеного техногенного навантаження. General Energy Institute of the National Academy of Sciences of Ukraine 2025-08-26 Article Article application/pdf https://systemre.org/index.php/journal/article/view/908 10.15407/srenergy2025.03.081 System Research in Energy; No. 3 (83) (2025): System Research in Energy; 81-91 Системні дослідження в енергетиці; № 3 (83) (2025): Системні дослідження в енергетиці; 81-91 2786-7102 2786-7633 en https://systemre.org/index.php/journal/article/view/908/813 Copyright (c) 2025 Oleksandr Kulykivskyi, Oleksandr Ponomarenko https://creativecommons.org/publicdomain/zero/1.0 |
| spellingShingle | monitoring intelligent systems sensor networks cloud technologies machine learning forecasting. Kulykivskyi, Oleksandr Ponomarenko, Oleksandr COMPARATIVE ANALYSIS OF METHODS AND TOOLS FOR AMBIENT AIR QUALITY MONITORING |
| title | COMPARATIVE ANALYSIS OF METHODS AND TOOLS FOR AMBIENT AIR QUALITY MONITORING |
| title_alt | Порівняльний аналіз методів та засобів моніторингу стану атмосферного повітря |
| title_full | COMPARATIVE ANALYSIS OF METHODS AND TOOLS FOR AMBIENT AIR QUALITY MONITORING |
| title_fullStr | COMPARATIVE ANALYSIS OF METHODS AND TOOLS FOR AMBIENT AIR QUALITY MONITORING |
| title_full_unstemmed | COMPARATIVE ANALYSIS OF METHODS AND TOOLS FOR AMBIENT AIR QUALITY MONITORING |
| title_short | COMPARATIVE ANALYSIS OF METHODS AND TOOLS FOR AMBIENT AIR QUALITY MONITORING |
| title_sort | comparative analysis of methods and tools for ambient air quality monitoring |
| topic | monitoring intelligent systems sensor networks cloud technologies machine learning forecasting. |
| topic_facet | monitoring intelligent systems sensor networks cloud technologies machine learning forecasting. моніторинг інтелектуальні системи сенсорні мережі хмарні технології машинне навчання прогнозування. |
| url | https://systemre.org/index.php/journal/article/view/908 |
| work_keys_str_mv | AT kulykivskyioleksandr comparativeanalysisofmethodsandtoolsforambientairqualitymonitoring AT ponomarenkooleksandr comparativeanalysisofmethodsandtoolsforambientairqualitymonitoring AT kulykivskyioleksandr porívnâlʹnijanalízmetodívtazasobívmonítoringustanuatmosfernogopovítrâ AT ponomarenkooleksandr porívnâlʹnijanalízmetodívtazasobívmonítoringustanuatmosfernogopovítrâ |