基于有效度和相对熵的短期电力负荷组合预测方法

Combination forecast method of short-term power load based on forecasting availability and relative entropy

负荷的精确预测是保证电力系统稳定调度的基础,为了更好地综合利用各单项预测方法,提高预测效果,该文引入相对熵组合方法对电力负荷进行预测。首先引入有效度的概念,对单项预测方法的准确程度进行判别,筛选出结果准确的多个单项预测方法,分别进行虚拟预测;并利用支持向量矩阵求解相对熵最值,得到动态的最优模型权重值,进行组合预测。以青海某区域实际负荷功率数据为例,利用ARIMA模型、BP神经网络、RBF神经网络及支持向量机(SVM)4种方法的预测结果进行相对熵组合预测,该预测方法与现有的组合预测方法对比,具有优越性与可行性。

Accurate forecasting of power load is the foundation of power dispatching and planning. In order to comprehensively utilize the existing single prediction method,this paper proposed a relative entropy method to predict power load.Firstly,the concept of forecasting availability was introduced to j udge the accuracy of each single prediction method,which can ensure the forecast result against the influence of some relatively inaccurate methods.Secondly virtually forecasting the remaining qualified method,the optimum value of relative entropy was solved by using sup-port vector matrix,and then dynamic weight value can be obtained.Taking the power data of a certain domain in Qinghai Province as an example,the forecast results of ARIMA model,BPNN model,RBFNN model and SVM were combinated by using relative entropy.Comparing to other combination methods,the proposed method is with availability and superiority.