Transformer–Gate Recurrent Unit-Based Hourly Purified Natural Gas Prediction Algorithm

• DOI: 10.3390/pr13010116
• Publication Date: 2025-01-06T09:37:10Z
• Authors: Chang Su; Jingcai Huang; Shasha Dong; Yuqi He; Ji Li; Luyao Hu; Xiao Liu; Yong Liao

With the rapid development of industrial automation and intelligence, consumption resources environmental impact production processes cannot today be ignored. Today, natural gas, as a commonly used energy source, produces significantly lower emissions carbon dioxide, sulphur nitrogen oxides from combustion than coal oil, can further purified to remove small amount impurities it contains, such compounds. Therefore, gas (hereinafter referred gas), clean plays an important role in realising sustainable development. At same time, It becomes more dispatch reasonably accurately, paramount factor is that load needs predicted accurately. this paper proposes Transformer–GRU-based hourly prediction model for gas. The uses Transformer data fusion feature extraction, then combines time series processing capability Gate Recurrent Unit (GRU) capture long-term dependencies short-term dynamic changes data. In paper, Chongqing Municipality 2020 was first preprocessed, divided into daily datasets according measurement step. Meanwhile, considering influence temperature factor, experimental dataset subdivided whether includes or not, Transformer–GRU built prediction, respectively. results show that, compared with Dual-Stage Attention-Based Neural Network (DA-RNN) GRU models alone, exhibits good performance terms coefficient determination, average absolute percentage error, mean square which well meet requirement accuracy has greater application value.