In this paper, we formally discuss a computational scheme, which combines a local weighted regression model with fuzzy transform (or F-transform for short). The latter acts as a reduction technique on the cardinality of the learning problem, resulting in a more efficient algorithm. We tested the proposed approach first through two typical benchmark problems, that is the Hénon and the Mackey–Glass chaotic time series, then we applied it to short-term forecasting problems. Short-term forecasting is important in the energy field for the management of power systems and for energy trading. Hence, we considered two typical application examples in this field, that is wind power forecasting and load forecasting. Numerical results show the effectiveness of the proposed approach through a comparison against alternative techniques.

中文翻译：

---

将局部学习与模糊变换结合使用：在短期预测中的应用

在本文中，我们正式讨论了一种计算方案，该方案将局部加权回归模型与模糊变换（或简称F变换）相结合。后者充当了学习问题的基数的简化技术，从而产生了更有效的算法。我们首先通过两个典型的基准问题（即Hénon和Mackey-Glass混沌时间序列）对提出的方法进行了测试，然后将其应用于短期预测问题。短期预测在能源领域对于电力系统的管理和能源交易非常重要。因此，我们考虑了该领域的两个典型应用示例，即风电预测和负荷预测。数值结果通过与替代技术的比较表明了该方法的有效性。