基于EMD与粗糙集及神经网络相结合的短期负荷预测

为了提高预测具有周期性和随机性的电力负荷精度,提出了一种基于经验模式分析(EMD)与粗糙集及神经网络相结合的短期负荷预测方法.该方法利用EMD的自适应性,自动地将目标负荷序列分解为若干个独立的内在模式分量.考虑影响电力负荷的气象因子和模式分量信息量较大,利用粗糙集进行了属性约简,约简后的各个分量采用相匹配BP神经网络模型分别进行预测,然后,相加各分量预测值得到最终预测结果.仿真试验表明,该方法与EMD - BP模型预测方法相比,具有较高的精度和较强的适应能力.

Short-term load forecasting based on EMD and the combination of rough set and neural network

In order to improve precision of load forecasting with periodicity and randomicity,this paper proposes short-term load forecasting based on EMD and the combination of rough set and neural network,a method which decomposes automatically the target load sequence into several independent IMF making use of EMD adaptability.Taking the large information content of meteorological factor and mode function influencing electrical load into consideration,based on rough set,the paper makes attributes reduction and after which adopts matching BP neural network mode to separately predict every function,totaling the predictive values of which to get the ultimate result.Simulation experiment shows that this method enjoys higher precision and stronger adaptability compared with EMD-BP mode.