Application of artificial intelligent maintenance of hydrogen systems in a Smart City

Abstract

Smart cities are technologically advanced urban areas that incorporate the internet of things (IoT) to improve the way we live in a sustainable manner. One of such use is in the management of clean energy use and storage. Hydrogen has been identified as a good solution for long term storage of energy and can be produced with the aid of electrolysers which use electricity such as from solar or wind systems to split water into hydrogen and oxygen gas. To ensure electrolysers work effectively there is a need to monitor their operation by taking various data using an IoT system and analyzing them to determine potential issues. This paper has developed a hybrid artificial intelligence concept comprising a deep reinforcement learning (DRL) and long short-term memory network (LSTM) for the intelligent maintenance of electrolysers. The DRL algorithm searches for the best data among others in an electrolyser with the highest correlation to a critical one which in this study is the hydrogen temperature. The DRL identified the cooling water temperature as having the highest correlation coefficient of 0.99. This data is then fed into the LSTM to predict the hydrogen temperature with a root-mean-squared error of 0.1351. The predicted sensor values are then used to control or shut down the electrolyser in the event of failure of the actual sensor.