The increase in CO2 emissions has led to a series of environmental problems, including global warming, making it imperative to reduce CO2 emissions. Adsorption carbon capture technologies have been widely researched, but there is currently a lack of comprehensive research on the optimization of cyclic performance that considers multiple objectives. This paper focuses on temperature swing adsorption (TSA) and develops algorithms for cyclic performance optimization. It employs machine learning techniques to conduct multi-objective optimization of the cycle. The calculation time for the surrogate model is only 1/1000 of the TSA mathematical model. The results indicate that the surrogate model obtained through machine learning accurately represents the cyclic performance under different operating parameters. There is a competitive relationship between productivity and exergy efficiency throughout the cycle. Recovery rate and exergy efficiency exhibit a dual relationship, both competitive and positively correlated. Purity and recovery rate show a purely positive relationship.
Keywords CO2, TSA, cycle performance, cycle optimization, cycle decision-making