科研之友
微信
新浪微博
Facebook
分享链接
摘要
Energy, especially oil, plays a special and irreplaceable role in the economic development, modern civilization, and social progress. With the rapid growth over the past few decades, China has gradually become a big power in oil consumption. In order to solve the contradiction between supply and demand as well as minimize social costs, it is necessary and useful to forecast the trend of China's oil consumption. In addition, reasonable and effective oil production and scheduling through forecasting progress also makes an important impact on the health economic development, social stability, and sustainable development. However, it is a challenging task to carry out such a forecasting because oil consumption is influenced by a number of factors, such as technology development, economic level, government policy, natural disaster, unexpected politic events, and so on. Therefore, it is difficult to forecast such a complex system with a single traditional model. This paper proposes an improved hybrid method known as GGNN, which combined the grey models, the back propagation (BP) neural network, and the genetic algorithm (GA) to take the advantages of linear model, nonlinear model, and swarm intelligence optimization, respectively. GM (1, 1) and improved grey models including an unbiased GM (1, 1), initial correction GM (1, 1), p value GM (1, 1), and background value GM (1, 1) are applied to capture the linear information. BP is widely used due to its nonlinear mapping capability; in this paper, it is used to capture the nonlinear information. Moreover, the GA is also applied to obtain the optimum weights and thresholds of the GGNN which is made up of all grey models and BP neural model. The superiority of this proposed method is examined by using the historical data of China's oil consumption. Assessment results demonstrate that the proposed method GGNN can improve the forecasting accuracy compared with some other existing methods.
- 出版日期2016-3
- 单位兰州大学
