基于EEMD与LSSVR的能源消费量多尺度预测——以广东省为例

发布时间：2018-07-01 21:08

本文选题：能源消费量 + 多尺度预测　；参考：《内蒙古大学学报(自然科学版)》2015年03期

【摘要】：由于能源消费内在的复杂性,传统的单尺度预测方法很难获得理想的预测效果.为提高能源消费量预测精度,提出了基于集合经验模态分解(EEMD)与最小二乘支持向量回归(LSSVR)的能源消费量多尺度预测模型.首先应用EEMD算法将能源消费量环比指数从高频到低频分解成若干结构更简单、变化更平稳、规律性更强、更易于预测的内在模态函数(IMF)和一个残差项;其次利用LSSVR对各IMF和残差项进行预测,并采用粒子群算法(PSO)选择最优的模型参数;然后将各分量的预测值直接加总求和重构出能源消费量环比指数的预测序列;最后通过逆环比化处理,获得原始能源消费量的最终预测值.利用该模型对1980-2013年广东省能源消费量进行实证分析,结果表明多尺度预测模型的确能够显著提高预测精度.
[Abstract]:Because of the inherent complexity of energy consumption, it is difficult for the traditional single-scale prediction methods to obtain ideal prediction results. In order to improve the prediction accuracy of energy consumption, a multi-scale energy consumption prediction model based on set empirical mode decomposition (EEMD) and least squares support vector regression (LSSVR) was proposed. Firstly, the EEMD algorithm is used to decompose the energy consumption index from high frequency to low frequency into a number of internal modal functions (IMF) and a residual term which are simpler, more stable, more regular, more predictable and easier to predict. Secondly, LSSVR is used to predict the IMF and residual terms, and Particle Swarm Optimization (PSO) algorithm is used to select the optimal model parameters, and then the prediction sequence of the energy consumption ring index is reconstructed by the sum of the predicted values of each component directly. Finally, the final prediction value of the original energy consumption is obtained by inverse loop processing. The model is used to analyze the energy consumption in Guangdong Province from 1980 to 2013. The results show that the multi-scale forecasting model can improve the prediction accuracy significantly.
【作者单位】：暨南大学经济学院;五邑大学经济管理学院;暨南大学管理学院;
【基金】：国家自然科学基金(71201010;71303174) 广东省自然科学基金(S2011010001591)
【分类号】：F426.2

【共引文献】