Research on urban gas load forecasting based on Bi-GRU model

Inhalt:
The prospective prediction of urban gas users' gas load is of great significance to guarantee the city's stable, safe and reliable gas use. Aiming at the difficulty of gas load prediction and the failure of LSTM network model to make full use of gas load data information. In this paper, Bi-GRU model is firstly constructed for feature learning. Secondly, multiple Bi-GRU network structures are superimposed together to build Stacked-Bi-GRU structure. Finally, the prediction of future gas load value is realized. The experimental results show that the gas load model proposed in this paper has higher prediction accuracy, which verifies the effectiveness of Bi-GRU model in the research field of gas load prediction.