Experimental implementation of a new control approach using an inverse neural network to on-demand hydrogen production. (December 2020)

• Record Type: Journal Article
• Authors: Cervantes-Bobadilla, M.; García-Morales, J.; Escobar-Jiménez, R.F.; Hernández-Pérez, J.A.; Gómez-Aguilar, J.F.; Olivares-Peregrino, V.H
• Is Part Of: Control engineering practice. Volume 105(2020)
• Publication Date: 2020-12
• Subjects: Hydrogen production control -- ANN -- ICANNi -- ANNi -- On-demand hydrogen production; Automatic control -- Periodicals; 629.89
• DOI: 10.1016/j.conengprac.2020.104631
• Languages: English
• ISSNs: 0967-0661

Abstract

Abstract: In this work, a new control approach for regulating the production of hydrogen gas on-demand of an electrolytic reactor was developed and experimentally tested. The control approach is based on an inverse artificial neural network (ANNi) and an integral control law (IC). The production hydrogen gas on-demand aims to use it as an additive in an internal combustion engine (ICE). For regulating the hydrogen flow rate (controlled variable), the current feed to the electrolyzer is manipulated. Therefore, for estimating the current feed to the electrolyzer an ANNi was formulated from an artificial neural network (ANN). The ANN's architecture is three inputs (current feed, temperature, and electrolyzer pressure), one neuron in the hidden layer, and one output (hydrogen flow). The integral control law is implemented for reducing the error estimation between the setpoint and the ANNi estimation. The proposed control scheme (Inverse Artificial Neural Network with an Integral Control (ICANNi)) showed excellent reference tracking, obtaining a root mean square error (RMSE) of 3 . 56 × 1 0 − 8, an average establishment time of 1.92 s, and maximum overshoots of the current of 1.8 Ampere in the reference changes. Highlights: Experimental implementation of a new control approach for on-demand hydrogen production. An inverse artificial neural network (ANNi) is design based on an artificial neural network (ANN). The new control approach called (ICANNi) will be integrated by the ANNi, and an integral control law.