基于多源数据的自适应电力负荷预测方法

为满足电力规划部门的实际需求,并充分利用海量开源数据,提出一种基于开源大数据,自主整理数据并自适应选择预测模型的电力负荷预测方法,该方法通过收集海量数据并归类,筛选得到多个与负荷预测强相关的数据源,并提出自适应负荷预测模型,该模型应用灰色预测函数、弹性系数预测函数、人均用电量预测函数、人工神经网络预测函数等多种数学方法,且可以根据数据来源进行相应拓展,并采用四种评价指标对多源预测结果进行修正。实例应用结果表明,该方法可以提高预测精度,工程实用价值较大。

Adaptive Power Load Forecasting Method Based on Multi source Data

To meet the actual needs of power planning departments and make full use of the massive open source data, this paper proposes a new method of load forecasting based on open source data, which can sort data autonomously and select prediction model adaptively. Firstly, this method collects and classifies mass data actively, and selects a number of data sources which are strongly correlated with load forecasting. Then adaptive load prediction model is proposed. The model can apply grey prediction function, elasticity coefficient forecast function, per capita consumption prediction function, artificial neural network prediction function and other mathematical methods, and can be expanded according to the corresponding data source. Considering multi source data prediction results on the basis of single method prediction, four kinds of evaluation indexes are used to correct the prediction results. The results of case analysis show that this method can realize the effective application of massive open source data and improve the prediction accuracy.