【摘要】 根据电网负荷混沌性的特点,提出一种基于模糊支持向量的核回归方法进行电力系统的负荷预测。同时提出多参数同步优化策略,增强了该方法的实用性和有效性。从理论上分析了小样本条件下,可以有效避免过学习的原因,该方法不需设计网络结构,降低了对实验人员经验的依赖程度。选取实际负荷时间序列数据,通过与神经网络法进行对比实验,结果显示出该方法的优越性和适用性,具有较好的实用价值和应用前景。更多还原

【Abstract】 According to the chaotic characteristic of power load,fuzzy support vector based kernel regression method is proposed for load forecasting.Then,multi-parameter synchronous optimization strategy is presented to speed up the optimization process.Case study shows that the method has lower error and can avoid over-fitting effectively,which is better than conventional neural network method.It is of great value for engineering application.更多还原