Integration of Artificial Neural Networks and Genetic Algorithm to Predict Oil Demand in China

X. Dong and S. Wang (PRC)


Oil demand, prediction, artificial neural networks, genetic algorithm


In this work, artificial neural networks based on a genetic algorithm (ANN-GA) are developed for predicting China’s oil demand in future. The back-propagation neural network is used to construct the relationship between gross domestic product, population, oil price, the numbers of motor vehicles and oil demand in China. The number of neurons in the hidden layer, the momentum and the learning rates of back propagation algorithm are determined using the GA algorithm. It is verified that genetic algorithm could find the optimal architecture and parameters of the back propagation algorithm. The ANN GA is tested and the results indicate that the China’s oil demand can be efficiently forecasted by this model. In addition, under simple estimation about socio-economic and oil use related indicators, the ANN-GA model is used to predict China’s oil demand in 2008-2020 and the results show that the growth rate of oil demand averagely is about 3.1% annually which is lower than Energy Information Administration (EIA) and International Energy Agency (IEA). This may be due to the impacts of improved energy efficiency, alternate energy sources and unexpected shocks to economy in the past three decades.

Important Links:

Go Back