基于负荷误差和经济发展趋势的组合预测模型在中长期负荷预测中的应用

The application of combination forecasting model in medium-long term load forecasting based on load error and economic development trend

针对组合预测未来预测期权重的确定,充分考虑电力需求与未来经济发展的关系,提出基于两个评价指标来计算各单一预测模型的权重,指标一是各单一模型预测值与实际负荷的误差,指标二是各预测值的年增长率与国内生产总值年增长率的误差。采用客观熵权法和主观G1法来确定两指标的相对重要性,最后综合得出各单一预测模型的权重。该改进组合预测模型解决了未来预测阶段权重的求解问题,通过对农网中长期电力负荷进行预测,结果表明该方法比常规方差优选组合预测模型更为准确和可信。

In point of the acquiring of the weight of the combination forecasting method in the future,a ndfully considering the relationship between power demand and economy development,this paper proposes two indicators to calculate the weight of each single model. O ne refers tothe value of errors between each single-model predictions and the actual load,andthe other refers to the errors between annual growth rate of each forecast value and annual growth rate of GDP. Theobjective entropy method and subjective G1 method are adopted to determine the relative importance of the two indicators and to integrately get the weight of each single model.The improved combination forecasting model sloves the weight solution in future forecast and is applied in medium-long term load forecasting for rural distribution network and the results show that this method is more accurate and credible than the conventional variance optimized combination forecasting model.