• Selvakumar T
• Dr. S.Somasundaram

Gas leakage and fire incidents originating from domestic and industrial Liquefied Petroleum Gas (LPG) installations remain a persistent cause of injury, property damage, and loss of life. Conventional safety devices typically rely on a single gas sensor and a fixed alarm threshold, which makes them prone to false alarms, insensitive to cylinder depletion, and incapable of autonomous corrective action. This thesis presents the design, development, and experimental evaluation of an AI-enabled automated LPG gas leakage and fire detection system that fuses multiple sensing modalities with a rule-based decision-making framework to provide reliable hazard detection, autonomous safety actuation, and real-time remote notification. The proposed system employs an ESP32 microcontroller as the central processing unit, interfaced with an MQ-2/MQ-6 combustible gas sensor for leakage detection, an infrared flame sensor supplemented by camera-based image-processing verification (using Python and OpenCV) for reliable fire detection, and a Force Sensitive Resistor (FSR) for continuous monitoring of gas cylinder load. Sensor readings are evaluated against a predefined rule set that fuses multiple parameters before triggering an automated response, comprising servo-actuated closure of the gas cylinder valve and relay-controlled activation of an exhaust fan. Combining a flame-sensor trigger with camera-based image verification substantially reduces false fire alarms caused by transient light sources, a limitation widely reported in single-sensor fire detectors. Real-time alerts for gas leakage, confirmed fire, and low cylinder weight are dispatched to the user through a Telegram bot, enabling remote awareness and timely intervention irrespective of the user's location. A working prototype was designed, assembled, and subjected to structured bench-level and system-level testing, including controlled gas-release trials, flame and light-interference tests, cylinder-depletion simulations, and full end-to-end integration runs. Experimental evaluation demonstrated reliable detection performance, sub-3-second detection-to-actuation response, and a substantial reduction in false fire triggers when compared with a single-sensor baseline. The results confirm that a carefully engineered rule-based multi-sensor fusion approach, augmented with lightweight image verification, can deliver dependable, low-cost, and field-deployable hazard protection without the computational overhead of a full machine-learning pipeline. The thesis concludes with a discussion of the system's contributions, its limitations, and directions for future enhancement, including on-device TinyML inference, fuzzy graded response, and multi-cylinder monitoring

LPG gas leakage detection, fire detection, rule-based artificial intelligence, sensor fusion, Force Sensitive Resistor, image processing, ESP32, Telegram bot, real-time monitoring, smart safety controls.

"Design and Development of Ai enabled Automated lpg gas leakage and fire detection system with real time monitoring and smart safety controls", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 9, page no.a681-a686, September-2026, Available :http://www.jetir.org/papers/JETIR2609084.pdf

"Design and Development of Ai enabled Automated lpg gas leakage and fire detection system with real time monitoring and smart safety controls", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 9, page no. ppa681-a686, September-2026, Available at : http://www.jetir.org/papers/JETIR2609084.pdf