中文摘要：

针对支持向量机在中长期负荷预测中关键参数选择的问题,引入蛙跳算法（SFLA）以优化基于支持向量机的中长期负荷预测算法,解决支持向量机参数选择问题。以对中国能源消费总量预测为例,对本文提出的改进算法进行验证。以1979—1999年的能源消耗量作为样本,对2000—2009年能量消耗量进行检验。研究结果表明：引入蛙跳算法后,与用粒子群（PSO）算法改进的支持向量机以及普通支持向量机方法相比,改进支持向量机预测精度分别提高2.34%和3.21%,算法运行时间分别增加51 s和109 s。

英文摘要：

By introducing the shuffled frog leaping algorithm to solve the random-choice problem of the key parameters,the new support vector machine optimized was proposed.In order to verify the optimized effect,based on the training exampling energy demand from 1979 to 1999,the improved SVM was used to forecast the total energy demand of China in 2000—2009.The results show that compared to the normal PSO-SVM and SVM responsively,the time of the improved SVM increases by 51 s and 109 s,and the precision of the proposed method increases by 2.34% and 3.21%,respectively.