Hydrogen leakage traceability is a key technology to ensure the safety and stability of hydrogen in the whole process of production, storage, transportation and usage. Traditional machine learning methods require manual processing of data features, which makes it difficult to cope with the high-dimensional features of multi-source hydrogen sensor data, resulting in complex model adjustments and a lack of generalization. Therefore, this study proposed a hydrogen leak tracing method based on Residual Neural Networks to improve the accuracy and efficiency of hydrogen leak tracing. Firstly, the multi-source hydrogen sensor data is combined into gray level maps and pre-processed. Then, the Residual Neural Networks model is constructed as the backbone network to adaptively extract the image features of hydrogen concentration and perform the traceability task. Finally, the effectiveness and advancement of the proposed method are tested on actual sensor signals. The comparative experimental results show that compared with other models, the Residual Neural Networks model exhibits better generalization and accuracy, with an average F1 score of 87.1%.
Wang et al. (2026) studied this question.