改进相似日评价函数在短期负荷预测中的应用

为提高短期负荷预测的精度,在分析传统相似日选取不足之处后提出了结合形状相似度与日特征相似度的两种改进的相似日评价函数,对其误差特性进行实例对比分析后得出了相似日方法在短期预测中部分共有的性质,最后结合回归预测方法,组成一种系统的短期负荷预测方法。通过实际数据的仿真实验证明该方法在传统相似日的基础上可以提高精度1.5%,是一种相对完整而有效的短期负荷预测方法。

Application of the Functions of Two Improved Similar Days in Short-term Load Forecasting

Traditional methods of similar days are of low forecasting accuracy. Based on the analysis of traditional methods of similar days, two methods to improve similar days are developed in this paper, which combine shape-similar days with day-character-similar days. In the mean time, the error character of two different improved methods is compared. Then the common character of the methods to improve similar days is gained. In the end, the method of improved similar days is combined with regression and they form a systematic method of short-term load forecasting. Examined by the emulator which is based on the actual data, the forecasting accuracy of the improved similar days in creases by 1.5 percent, which is an effective and comprehensive method of short-term load forecasting relatively.