基于预测决策的组合短期负荷预测

A Combination Short-Term Load Forecasting Based on Decision-making

以提高短期负荷预测精度为目的,首先选择五种预测模型分别对算例进行虚拟负荷预测,再利用决策理论知识,通过几率矩阵法来筛选预测模型,舍弃不符合要求的模型,最后以各时段残差平方和最小为目标函数确定各个模型权重系数。使用选中的模型进行组合短期负荷预测,并将预测结果与各单一模型的预测结果进行比较。说明了预测决策的可参考特性和实用特性,也说明了组合预测模型的拟合精度优于各种单一模型,得到了较好的预测效果。

In order to improve the short-term load forecasting accuracy, first the paper choosed five models to do virtual load forecasting for the same case respectively, then used the decision-making theory to sift prediction models through the odds matrix and abandon the models not complied with requirements. Finally it used the minimized residual sum of squares at each period as objective function to determine the weighting of selected models, and the selected models to do a combination short-term load forecasting and compared the result with the single one. Its testified decision-making was referable and practical, the fitting accuracy of combination load forecasting was better than every single model.