Oil Demand Forecasting from The Perspective of Alternatives

In recent years, events such as oil sanctions, the Russia-Ukraine conflict, and the COVID-19 pandemic have led to global instability and extreme fluctuations in international oil prices. This has slowed the growth of oil consumption, while the development of new energy technologies and the introduction of alternative energy sources are gradually transforming China’s energy demand structure. Accurately forecasting oil demand has thus become crucial for national energy security and economic development. This paper predicts China’s future oil demand by examining commonly used energy sources and focusing on oil substitutes. Sixteen relevant factors were selected, and machine learning and neural network methods were applied to assess their impact on China's oil consumption demand. Experimental results demonstrate that considering oil substitutes in variable selection is feasible for predicting China’s oil demand. The combined LSTM and attention mechanism approach achieved the highest prediction accuracy, stability, and overall performance. The findings of this empirical study offer valuable insights for future research utilizing deep learning techniques to forecast oil demand.