Back propagation neural network based proportional-integral hybrid control strategy for a solar methane reforming reactor

• Published in: International journal of hydrogen energy 2024-01, Vol.49, p.1258-1271
• Main Authors: Zhang, Kai-Ran, Tang, Xin-Yuan, Yang, Wei-Wei, Li, Jia-Chen, Zhang, Ren-Zhong
• Format: Article
• Language: English
• Subjects: Active regulation; Hybrid control strategy; Methane steam reforming; Neural network; Solar thermochemical reactor
• ISSN: 0360-3199; 1879-3487
• DOI: 10.1016/j.ijhydene.2023.09.215

To handle the impact of solar radiation fluctuations on the solar reactor, a new proportional-integral (PI) hybrid control strategy based on back propagation neural network (BPNN) is designed in this study to achieve stable operation. Firstly, the BPNN model optimized by genetic algorithm is trained using reactor model simulation data. Then, the BPNN model is combined with PI feedback control to obtain a BPNN-based PI feedforward-feedback hybrid control strategy for regulating the inlet flow rate. Finally, the effectiveness of the control strategy during transient operation under different disturbances is tested. The results show that during actual solar radiation disturbance, the integral square errors (ISE) of outlet hydrogen fraction and methane conversion for BPNN-based PI hybrid control are 7.03 × 10−6 and 1.05 × 10−4, respectively. Compared to traditional PI feedback control, the new hybrid control strategy represents a reduction of 94% and 91% in ISE, indicating that the strategy can significantly improve production stability. •BPNN based PI hybrid control strategy is designed for solar methane reformer.•Different control strategies are examined with various solar fluctuation conditions.•Stabilities of outlet hydrogen fraction and methane conversion are clearly improved.•The proposed BPNN based PI hybrid control strategy reduces ISE by 90%.•Dynamic tracking performance of proposed control strategy is revealed.