Skip navigation


  • DSpace logo
  • Home
  • Collections
  • Researcher Profile
  • Explore by
    • Researcher Profile
  • VGU library
  • Help
  • User Guide
  • Sign on to:
    • My DSpace
    • Receive email
      updates
    • Edit Account details

VGU RESEARCH REPOSITORY


Please use this identifier to cite or link to this item: https://epub.vgu.edu.vn/handle/dlibvgu/2129
DC FieldValueLanguage
dc.contributor.authorLe Vu Tuan Anhen_US
dc.date.accessioned2026-01-08T08:51:31Z-
dc.date.available2026-01-08T08:51:31Z-
dc.date.issued2025-
dc.identifier.urihttps://epub.vgu.edu.vn/handle/dlibvgu/2129-
dc.description.abstractTraditional fire detection systems suffer from a critical trade-off between sensitivity and reliability, leading to high false alarm rates and the risk of delayed detection. This research addresses this challenge by developing an integrated system that improves fire prediction accuracy while significantly reducing nuisance false alarms. The proposed system enables reliable early fire detection alongside real-time, localized Air Quality Index (AQI) classification through a custom hardware platform equipped with a multi-sensor array (BME680, PMS7003, CCS811, and MiCS-4514) to capture comprehensive environmental time-series data. A labeled dataset was constructed by simulating incipient fire scenarios, common nuisance sources such as cooking fumes and steam, and normal environmental conditions. The core methodology employs a two-stage hybrid model in which a stacked LSTM autoencoder performs self-supervised representation learning to extract robust temporal features, which are subsequently used as high-quality inputs for a supervised feedforward neural network trained to accurately classify events into Fire, Nuisance, and No Fire categories, thereby reducing nuisance-based false alarms. In parallel, the framework incorporates classification models that map multi-sensor data to standard AQI categories. By integrating fire detection and air quality monitoring into a unified system and leveraging self-supervised learning to enhance supervised classification performance, this research delivers an advanced environmental monitoring solution that improves public safety through earlier fire detection, substantial false alarm reduction, and immediate hazard assessment.en_US
dc.language.isoenen_US
dc.rightsAttribution-NonCommercial 4.0 International*
dc.rights.urihttps://creativecommons.org/licenses/by-nc/4.0/*
dc.subjectMulti-sensor systemsen_US
dc.subjectDeep learningen_US
dc.subjectAir quality index (AQI)en_US
dc.subjectFalse alarm reductionen_US
dc.titleEnhanced fire detection and air quality index classification using multi-sensor time series analysis with stacked LSTM autoencodersen_US
dc.typeThesisen_US
item.fulltextWith Fulltext-
item.languageiso639-1other-
item.grantfulltextrestricted-
Appears in Collections:Mechatronics & Sensor Systems Technology (MSST)
Files in This Item:
File Description SizeFormat Existing users please Login
Enhanced fire detection and air quality index classification using multi-sensor time series analysis with stacked LSTM autoencoders.pdf5.31 MBAdobe PDF
Show simple item record

Google ScholarTM

Check


This item is licensed under a Creative Commons License Creative Commons

© Copyright 2020 by Vietnamese - German University Library.
Add: Ring road 4, Quarter 4, Thoi Hoa Ward, Ben Cat City, Binh Duong Province
Tel.:(0274) 222 0990. Ext.: 70206