【摘要】 分析了电力系统负荷预测目前采用方法的不足,并根据电网负荷的特点提出一种基于模糊支持向量的核回归方法,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,该方法可以有效避免过学习的原因,它不需设计网络结构,降低了对实验人员经验的依赖程度.与神经网络法进行对比实验,实验结果表明了该方法的优越性和适用性,并具有较好的实用价值和应用前景.更多还原

【Abstract】 The shortcomings of current used methods in load forecasting of power systems.A fuzzy support vector kernel regression method is presented.A multi-parametes synchronous optimization is also presented to enhance the effectivemess of the method.Experiments based on power system data show that this kind of kernel regression method brings lower error,can avoid over fitting effectively,no need to design nueral net-structure,and use fewer time than artificial nueral network.It seems to be the most promising one and has some more applied value.更多还原