Gas Sensor Based Remote Environmental Monitoring : Wildfire Detection
Shahzaib, Mohammad (2023)
2023
Master's Programme in Information Technology
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2023-05-10
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202304274786
https://urn.fi/URN:NBN:fi:tuni-202304274786
Tiivistelmä
Environmental monitoring has become increasingly important as it involves the use of sensors to assess the state of the ambient environment, particularly in the context of climate change and its impacts. IoT technologies play a critical role in collecting real-time data from environmental sensing nodes, enabling timely decision-making and resource allocation.
The goal of environmental monitoring is to establish a mesh network of sensing devices across extensive forest regions, allowing the automatic collection of data samples for various purposes such as wildfire monitoring, habitat conservation, and pollution tracking. By integrating machine learning algorithms with gas sensor data, early warning systems can be developed to provide rapid alerts in case of wildfires, potentially saving lives and valuable resources.
Remote environmental monitoring also supports forest management through weather prediction, outdoor asset monitoring, and the assessment of ecological changes. This holistic approach to environmental monitoring contributes to sustainable forest management practices and helps mitigate the risks posed by climate change.
Between 2011 and 2021, wildfires have become an increasingly concerning issue, with an average of 62,805 wildfires occurring annually and affecting approximately 7.5 million acres. Traditional fire detection systems have proven to be unsuitable for outdoor environments, as they lack real-time detection capabilities and are not optimized for large-scale deployments.
This research aims to address these challenges by utilizing gas sensors combined with embedded devices to detect fire and smoke in outdoor areas using machine learning algorithms. The objective of this thesis is to apply chemical gas sensors in an E-Nose system to detect early-stage wildfires in forests, thereby minimizing their impact on ecosystems and human settlements. Fire and smoke detection are achieved through the assistance of machine learning algorithms that analyze data from the sensors. When fire or smoke is detected, the system promptly sends an alert to the relevant authorities, enabling a rapid response. The methodology involves collecting appropriate data sets, cleaning and preparing the data, training the models, and testing and validating their performance.
Artificial Neural Networks (ANN) and Machine Learning Classifiers are employed to develop prediction models for early-stage wildfire detection. These advanced techniques improve the accuracy and reliability of detection systems, ultimately contributing to better wildfire management and more sustainable forest ecosystems.
The goal of environmental monitoring is to establish a mesh network of sensing devices across extensive forest regions, allowing the automatic collection of data samples for various purposes such as wildfire monitoring, habitat conservation, and pollution tracking. By integrating machine learning algorithms with gas sensor data, early warning systems can be developed to provide rapid alerts in case of wildfires, potentially saving lives and valuable resources.
Remote environmental monitoring also supports forest management through weather prediction, outdoor asset monitoring, and the assessment of ecological changes. This holistic approach to environmental monitoring contributes to sustainable forest management practices and helps mitigate the risks posed by climate change.
Between 2011 and 2021, wildfires have become an increasingly concerning issue, with an average of 62,805 wildfires occurring annually and affecting approximately 7.5 million acres. Traditional fire detection systems have proven to be unsuitable for outdoor environments, as they lack real-time detection capabilities and are not optimized for large-scale deployments.
This research aims to address these challenges by utilizing gas sensors combined with embedded devices to detect fire and smoke in outdoor areas using machine learning algorithms. The objective of this thesis is to apply chemical gas sensors in an E-Nose system to detect early-stage wildfires in forests, thereby minimizing their impact on ecosystems and human settlements. Fire and smoke detection are achieved through the assistance of machine learning algorithms that analyze data from the sensors. When fire or smoke is detected, the system promptly sends an alert to the relevant authorities, enabling a rapid response. The methodology involves collecting appropriate data sets, cleaning and preparing the data, training the models, and testing and validating their performance.
Artificial Neural Networks (ANN) and Machine Learning Classifiers are employed to develop prediction models for early-stage wildfire detection. These advanced techniques improve the accuracy and reliability of detection systems, ultimately contributing to better wildfire management and more sustainable forest ecosystems.