中国能源市场波动预警分析——基于自组织竞争神经网络

为了对中国能源市场波动进行预警分析,本文立足能源市场构成要素的角度,从能源生产、人均消费和能源进出口贸易等方面提取了7项指标作为基础数据,对数据进行归一化处理后建立关于中国能源市场的自组织竞争神经网络模型,并对1990-2009年中国能源市场的波动状况进行实证分析.以前15年的数据作为训练样本数据,后5年的数据作为预警测试数据,用{1,2,3}分别代表不同的波动类型,大幅波动期、相对稳定期和小幅波动期.结果发现,该模型得出的结果与现实情况相符,1990-1994年为一个小幅波动期,1996-2000年之间处于一个相对稳定的波动时期,从2001-2009年我国能源市场表现的非常活跃,对能源的需求仍然很大.通过该方法对掌握市场波动规律,预警能源市场波动具有一定的指导意义.

Analysis of China's Energy Market Volatility Early-warning——Based on Self-organizing Competitive Neural Networks

This paper extracts seven indicators from energy production,per capita consumption and energy import and export trade in elements of energy market as the basic data for analyzing China’s energy market volatility,and establishes the self-organizing competitive neural networks model after normalizing the data,which empirically analyzes the volatility from 1990 to 2009.This paper takes the previous 15 years’ data as training data,recent 5 years’ data as warning test data,and represents the different wave type(period of sharply fluctuation,relatively stable and slight fluctuation) with {1,2,3}.It is found that the results of the model in line with reality.1990 to 1994 is a slight fluctuation period,and 1996 to 2000 in a relatively stable period of fluctuations.China’s energy market is very active from 2001 to 2009 and the energy demand is still enormous.It is significant to master the market rule and predict the energy market volatility.