改进模糊聚类负荷预测方法的探讨

针对传统的c-均值模糊聚类算法易陷入局部最优解、初始值c值的给定存在着很大的人为因素以及在整个计算过程中无法自我调节的缺陷,利用遗传算法的全局寻优能力并采用一种新式的双码染色体编码方法对传统的c-均值模糊聚类算法进行了改进,同时将这一自适应的SFGO（Sampling Fuzzy c-means with Genetic Optimization）算法运用到电力系统的中长期负荷预测中,得到了比较好的效果。

Discussion on Improving Fuzzy Clustering Algorithm to Power Load Forecasting

There are some drawbacks of classical fuzzy c-means as follow: First,easily to fall into the partly best solution.Second,the choice of the initiate c value is too much concerned with artificial factors.Third,it lacks of self-adaptive ability.As to compensate for these aforementioned drawbacks,a new algorithm named self-adaptive SFGO is improving in this essay.The new algorithm takes advantage of the superiority of GA in search of the total best solution,and develops a new coding method called double chromosome code.When it is employed to long-term power load forecasting,a better result can be accessed.