Real-Time Monitoring of Toxic Gases using Artificial Neural Networks with SnO2-based Thick Film Gas Sensors

Abstract

The present study aims to develop an advanced system for real-time identification and measurement of harmful gases using Pd-doped SnO2-based thick film gas sensors integrated with Artificial Neural Networks. These sensors, renowned for their exceptional sensitivity and selectivity, undergo testing in a meticulously controlled gas chamber environment. Here, gas concentrations, temperature, and humidity are meticulously controlled. The gas chamber setup allows for mixing gases like carbon monoxide, nitrogen dioxide, and sulfur dioxide with nitrogen gas to achieve desired concentrations ranging from to . Temperature is kept at and relative humidity is maintained between to . The sensitivity analysis of the sensor demonstrates its adeptness in detecting low concentrations of target gases, with sensitivity increasing as gas concentrations rise, also the selectivity assessments highlight their ability to accurately differentiate between target gases and common interferents, ensuring precise detection even in complex gas mixtures. Response time testing indicates rapid detection capabilities, crucial and useful for emergencies. The Artificial Neural Network model, trained via the backpropagation algorithm, demonstrates remarkable accuracy, precision, recall, and F1 scores in predicting gas concentrations. The findings indicate that this integrated system offers a reliable and efficient solution for toxic gas detection, with potential applications in industrial safety and environmental monitoring.