A novel combination of machine learning and intelligent optimization algorithm for modeling and optimization of green ammonia synthesis

Green ammonia synthesis is an important industrial chemical process, which is widely applied in fields such as fertilizers, petrochemicals and fuel cells. In order to improve green ammonia production and reduce energy consumption, this article focuses on a deeper understanding of the kinetic behavior of ammonia synthesis process system. To this end, a physics-informed sparse identification modeling and optimization framework for ammonia synthesis plant is proposed in this paper, which highlights in-depth exploration of reaction mechanisms, kinetic equations, and optimization methods. The proposed method can deal with the time series information generated by the complicated ammonia synthesis process system with noise. More importantly, the proposed method is found to have distinctive interpretability that from the parameters of differential equation governing the observable data can be deduced. A bald eagle search algorithm is used to solve the maximum yield problem of green ammonia, which can obtain the optimal reactor length and the maximum ammonia profit under physical limitation conditions. The simulation results illustrated that the proposed optimization method was highly competitive with other state-of-art global optimization methods. Finally, the effectiveness and robustness of the proposed method have been demonstrated on ammonia synthesis plant by achieving good and competitive model interpretation and accuracy.