Hydrogen Release Modelling for Analysis Using Data-driven Autoencoder with Convolutional Neural Networks

High-accuracy gas dispersion models are necessary for predicting hydrogen movement, and for reducing the damage caused by hydrogen release accidents in chemical processes. In urban areas, where obstacles are large and abundant, computational fluid dynamics (CFD) would be the best choice for simulating and analyzing scenarios of the accidental release of hydrogen. However, owing to the large computation time required for CFD simulation, it is inappropriate in emergencies and real-time alarm systems. In this study, a non-linear surrogate model based on deep learning is proposed. Deep convolutional layer data-driven autoencoder and batch normalized deep neural network is used to analyze the effects of wind speed, wind direction and release degree on hydrogen concentration in real-time. The typical parameters of hydrogen diffusion accidents at hydrogen refuelling stations were acquired by CFD numerical simulation approach, and a database of hydrogen diffusion accident parameters is established. By establishing an appropriate neural network structure and associated activation function, a deep learning framework is created, and then a deep learning model is constructed. The accuracy and timeliness of the model are assessed by comparing the results of the CFD simulation with those of the deep learning model. To develop a dynamic reconfiguration prediction model for the hydrogen refuelling station diffusion scenario, the algorithm is continuously enhanced and the model is improved. After training is finished, the model's prediction time is measured in seconds, which is 105 times quicker than field CFD simulations. The deep learning model of hydrogen release in hydrogen refuelling stations is established to realize timely and accurate prediction and simulation of accident consequences and provide decision-making suggestions for emergency rescue and personnel evacuation, which is of great significance for the protection of human life, health and property safety.