摘要
针对中期电力负荷预测,提出一种具有自适应噪声的完整集成经验模态分解(CEEMDAN)-排列熵和泄漏积分回声状态网络(LIESN)的组合预测方法。CEEMDAN方法在负荷序列分解的每一阶段添加特定的白噪声,通过计算唯一的余量信号以获取各个模态分量,与EEMD方法相比,其分解过程是完整的。为降低负荷非平稳性对预测精确度的影响以及减小计算规模,采用CEEMDAN-排列熵方法将负荷时间序列分解为具有复杂度差异的不同子序列,通过分析各个子序列的内在特性,分别构建相应的LIESN预测模型,最终对预测结果进行叠加。将该方法应用于不同地区的中期峰值电力负荷预测实例中,并与其他组合预测以及单一预测方法进行比较。实验结果表明,所提出的方法有很高的预测精确度,显示出其有效性和应用潜力。
Based on complete ensemble empirical mode decomposition with adaptive noise( CEEMDAN)-permutation entropy and echo state network with leaky integrator neurons ( LIESN) , a kind of combined forecasting method was proposed for medium-term power load forecasting. In the CEEMDAN method, a particular white noise was added at each stage of the decomposition and a unique residue was computed to obtain each intrinsic model function( IMF) , compared with EEMD, the resulting decomposition is com-plete. In order to weaken the influence of non-stationary effects of the load series on the prediction accu-racy and reduce computation scale, the load time series was decomposed into a series of subsequences with obvious differences in complex degree by using CEEMDAN-permutation entropy, and the corre-sponding LIESN forecasting model was built respectively by analyzing the inner characteristics of each subsequence. Simultaneously, the ultimate forecasting results can be obtained by the superposition of the corresponding forecasting model. The proposed method was applied to electricity peak load forecasting in-stances in different areas and compared with other combined and single forecasting methods. Experiment results confirm that the proposed method has a high prediction precision, and show the effectiveness and applicability.
出处
《电机与控制学报》
EI
CSCD
北大核心
2015年第8期70-80,共11页
Electric Machines and Control
基金
国家自然科学基金(51467008)
甘肃省高等学校基本科研业务费专项资金项目(620026)
关键词
负荷预测
组合模型
集成经验模态分解
回声状态网络
排列熵
load forecasting
combined model
ensemble empirical mode decomposition
echo state net-work
permutation entropy