摘要： 针对传统的中长期模糊聚类预测算法自变量权重选择不合理、截水平集合元素不全面、相关因子计算方法单一等缺陷，提出改进的预测算法。该算法利用关联度分析计算自变量权重，通过建立相关因子计算方法库，按照相对传递总偏差最小原则选择最佳相似矩阵进行聚类，以等价矩阵所有元素的去重集合作为截水平集合求最佳聚类。实验结果证明该算法可提高预测的准确性。

关键词: 模糊聚类, 相关因子, 相似矩阵, 关联度分析, 中长期电力负荷预测

Abstract: Classical fuzzy clustering algorithm has some drawbacks including that the computing of independent variable weights is unreasonable, the set of horizontal section members is slurred, the computational methods of correlation factor are single and so on. In order to solve the problems above, this paper proposes a new algorithm named improved fuzzy clustering algorithm. It uses association analysis to compute the independent variable weights, sets up a method warehouse and uses it to calculate the correlation factors, and selects distinct members of the equivalent matrix as the set of horizontal section. Experimental result demonstrates that the new algorithm increases the accuracy of forecast.

Key words: fuzzy clustering, correlation factor, similar matric, association analysis, forecast of middle and long term electric power load

中图分类号: 

• TP301.6