Adaptive Deep Learning Modeling of Green Ammonia Production Process Based on Two-Layer Attention Mechanism LSTM DOI Open Access
Jie Yang, Ji Zhao, Zhongbo Hu

et al.

Processes, Journal Year: 2025, Volume and Issue: 13(5), P. 1480 - 1480

Published: May 12, 2025

Green ammonia, as a zero-carbon energy carrier, has emerged core process for achieving transition and chemical industry decarbonization through renewable energy-powered electrolytic hydrogen production integrated with low-carbon Haber–Bosch ammonia synthesis. However, the strong coupling among multiple units in green systems, combined operational data characteristics of nonlinearity, uncertainty, noise interference, multi-timescale dynamics, creates significant challenges accurately predicting yields key indicators, ultimately hindering online parameter optimization restricting improvements efficiency effective carbon emission control. To address this, this study proposes dual-layer attention LSTM model. The architecture constructs two sequential mechanisms: first layer being an input mechanism screening critical followed by second temporal that dynamically captures time-varying feature weights, enabling adaptive analysis sub-window contribution discrepancies to output variables across time steps. Furthermore, model is implemented validated on simulation platform energy-coupled demonstration project, comparative analyses conducted against conventional other baseline models. Experimental results demonstrate proposed effectively adapts complex scenarios production, including fluctuating inputs reaction conditions, providing reliable support yield prediction optimization. developed methodology not only provides novel approach intelligent modeling systems but also establishes technical foundation digital twin-based real-time control dynamic scheduling research.

Language: Английский

Adaptive Deep Learning Modeling of Green Ammonia Production Process Based on Two-Layer Attention Mechanism LSTM DOI Open Access
Jie Yang, Ji Zhao, Zhongbo Hu

et al.

Processes, Journal Year: 2025, Volume and Issue: 13(5), P. 1480 - 1480

Published: May 12, 2025

Green ammonia, as a zero-carbon energy carrier, has emerged core process for achieving transition and chemical industry decarbonization through renewable energy-powered electrolytic hydrogen production integrated with low-carbon Haber–Bosch ammonia synthesis. However, the strong coupling among multiple units in green systems, combined operational data characteristics of nonlinearity, uncertainty, noise interference, multi-timescale dynamics, creates significant challenges accurately predicting yields key indicators, ultimately hindering online parameter optimization restricting improvements efficiency effective carbon emission control. To address this, this study proposes dual-layer attention LSTM model. The architecture constructs two sequential mechanisms: first layer being an input mechanism screening critical followed by second temporal that dynamically captures time-varying feature weights, enabling adaptive analysis sub-window contribution discrepancies to output variables across time steps. Furthermore, model is implemented validated on simulation platform energy-coupled demonstration project, comparative analyses conducted against conventional other baseline models. Experimental results demonstrate proposed effectively adapts complex scenarios production, including fluctuating inputs reaction conditions, providing reliable support yield prediction optimization. developed methodology not only provides novel approach intelligent modeling systems but also establishes technical foundation digital twin-based real-time control dynamic scheduling research.

Language: Английский

Citations

0