IntelliGas: An Intelligent Hazardous Gas Monitoring System

Abstract

The rising rate of urbanization and industrialization has significantly increased the probability of unintended gas emissions. Those emissions can be highly dangerous to affect humans, work zone safety and even the environment. Traditional approaches of gas monitoring applications are limited to alarming users about the presence of hazardous gases in real-time with negligible scope for predicting the events beforehand. In this paper, we present an Edge AI-based hazardous gas detection and prediction monitoring framework that allows real-time monitoring along with predicting the chances of any hazardous event using Machine Learning algorithms. MH-Z19E and MQ-series gas sensors connected to Raspberry Pi edge-device are used to sense the real-time levels of carbon dioxide (CO₂), liquefied petroleum gas (LPG) and smoke. The data fetched from sensors will be locally analyzed by implementing edge-computing and MQTT-integrated pipeline stores the sensor data to MongoDB database for long-term historical data analysis and model training. Built Machine Learning algorithms based on Random Forest, XGBoost, and Long Short-Term Memory (LSTM) are used for predicting future levels of sensed gas as well as identifying abnormal environmental signatures. The integrated approach of predicting gas levels based on historical data along with providing real-time monitoring alerts for gas sensing provides a fool-proof solution for early warning systems, abnormality detection, and safety management use-cases. Designed framework can be utilized for various industries for commercial as well as environmental monitoring purposes