微网用户短期负荷预测相似日选择算法

微网用户负荷基荷小、波动性和随机性大，增大了短期负荷预测难度。科学合理地选择相似日可以在一定程度上改善短期负荷预测的效果。分析了相似日选择的影响因素，针对微网用户负荷特点，提出了一种负荷点尺度上的相似日选取算法。该算法考虑了前几日气象因素的累积效应、短期负荷的连续性和周期性及时间距离的影响，其相似日评价函数计及了日特征相似和局部形相似，并引入时间因子，克服了传统人工经验选取相似日算法的主观性，使得选择的相似日更加客观合理。实例验证表明，该方法所选择的相似日用于微网用户短期预测时，可以提高预测精度，有一定的使用价值。

Table 2 Estimation accuracy of three forecasting methods Method for Selecting Similar Days in Short-term Load Forecasting of Microgrid

The small base load, high fluctuation and randomness of the microgrid increase the difficulty of short-term load forecasting. Scientific and proper selection of similar days can improve to some extent the effectiveness of the short-term load forecasting. Firstly, the factors which may affect selecting similar days are analyzed in this paper. Then, in view of the characteristics of microgrid load, a novel method for selecting similar days is proposed in microgrid short-time load forecasting based on load point scale. The proposed method considers the cumulative effects of the weather factors, the continuity and periodicity of short-term load and the effects of time distance, and its evaluation function of day character similarity takes into account of day similarity and partial similarity and introduces the time factor. Thus, the proposed method overcomes the subjectivity of the traditional method which is based on personal experience. Case study demonstrates that the proposed method can improve the accuracy of short-term forecasting and can be applied in practice.