基于相似度与神经网络的协同短期负荷预测模型

Coordinated short-term load forecasting based on similarity degree and ANN

为了考虑除负荷本身外的其他因素对短期负荷的影响,提出了基于相似度与神经网络的短期协同预测模型。该模型首先通过计算负荷曲线的相似度对历史数据进行排序,然后选择与预测时刻相似度较相近的数据对未来时刻的负荷利用相似度进行预测,对于出现的误差,通过神经网络结合其他因素进行预测纠正。实验结果证明,该协同预测模型较之单纯的BP神经网络预测模型具有较高的预测精度。

To make full consideration on the influences of factors other than the load itself on the short-term load, a coordinated short-term load forecasting model based on similarity degree and ANN was proposed. The load curve similarity was calculated to sort the history data at first, and the nearest history data were chosen for forecasting. In addi- tion, the Artificial Neutral Network （ANN） model was used to correct the errors. Example case showed that the model was more precise than the BP-ANN model.