Answers are grounded in this paper and cite their sources.
The concentration of gases in gas streams can be monitored using sensors. However, gas sensors can lose their response accuracy due to mechanical wear or damage, and environmental factors such as exposure to unusual temperature and pressure conditions. Therefore, it is paramount to design a hybrid artificial intelligence system to identify any sensor malfunction and the need to recalibrate or replace a sensor or a suite of sensors for real-time quantification of compositions of gas streams. Hence, this study proposes a hybrid artificial intelligence system for real-time monitoring of gas concentrations in a gas stream and recalibration of gas sensors. This system provides remote access for monitoring gas concentrations predicted by a machine learning model and sensor readings programmed in a wireless device or an application in a wireless device, enabling users to identify when certain set thresholds of gas concentrations are exceeded and to identify malfunction of sensors when predetermined deviations between the sensor readings and the machine learning predictions are exceeded, for quality control and assurance. The design also signals the need for recalibration or replacement of sensors, for more accurate readings. Therefore, this study developed a methodology for the design of the hybrid system and demonstrated the feasibility of operating the system in a nitrogen gas stream. Application of the system for carbon dioxide capture and storage was also explored. Machine learning models were developed for binary and multi-component gas mixtures using Python programming language. The findings of the study revealed that the error in quantification of gas concentrations for the binary gas mixture is less than the errors for the multi-component gas mixtures, using machine learning models. Therefore, while operating the hybrid artificial intelligence system for real-time quantification of impurities in gas streams, higher deviations in gas concentration between the sensors' readings and the machine learning model predictions should be allowable for the multi-component gas mixture compared to the binary gas mixture, as long as their set level of tolerance for the gas mixture is not exceeded.
No takes yet. Share an insight, caveat, or question.
Aminaho et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: