摘要
为提高短期电力负荷的预测精度,提出一种基于经验模态分解、计量经济学模型和神经网络混沌模型的组合预测方法。首先,利用经验模态分解将负荷序列分解成一系列本征模态函数及余项;其次,针对不同分量的特性,建立不同的模型进行预测;最后,将所有分量的预测值求和作为最终的预测结果。以美国宾夕法尼亚州–新泽西州–马里兰州(Pennsylvania-New Jersey-Maryland,PJM)电力市场为实例,验证了普通日负荷和特殊日负荷的预测效果,此外,将该方法与其他预测方法进行了比较,算例表明,该方法具有较高的预测精度。
To improve the accuracy of short-term load forecasting, a hybrid prediction method based on empirical mode decomposition （EMD）, econometric model and neural network chaotic model was proposed. Firstly, the load series was decomposed into several intrinsic mode function and remainder by EMD; then different models were built for forecasting according to the features of decomposed components; finally, the summation of all results forecasted by individual components was regarded as final forecasting result. Taking Pennsylvania-New Jersey-Maryland （PJM） electricity market for example, the proposed method was verified by forecasting results of ordinary day and that of special day. Calculation example results showed that the results forecasted by the proposed method were more accurate than those by other load forecasting methods.
出处
《电网技术》
EI
CSCD
北大核心
2011年第9期181-187,共7页
Power System Technology
基金
国家自然科学基金项目(70971038)
高等学校博士学科点专项科研基金资助项目(20070079005)~~
关键词
负荷预测
经验模态分解
计量经济学模型
神经网络混沌模型
load forecasting
empirical ,modedecomposition （EMD）
econometric model
neural networkchaotic model
