基于区间二型模糊逻辑的电力负荷预测研究  

Forecasting study of power load based on interval type-2 fuzzy logic method

作　　者：郑高 肖建[2] 

机构地区：[1]公安海警学院机电管理系,浙江宁波315801 [2]西南交通大学电气工程学院,四川成都610031

出　　处：《电机与控制学报》2012年第9期26-32,共7页Electric Machines and Control

基　　金：国家自然科学基金(60674057);中央高校基本科研业务费专项资金(SWJTU09ZT11)

摘　　要：针对电力负荷具有强随机性而难以准确预测的问题,引入二型模糊逻辑以减小预测误差。建立了区间二型模糊模型用于电力负荷的短期时间序列预测,先利用反向传播算法来调节模型的参数,再利用相似度来精简冗余模糊集合,最后利用奇异值分解法来优选模糊规则。同时,建立了基于反向传播算法的一型模糊模型与区间二型模糊模型,以及基于反向传播-奇异值分解混合迭代算法的区间二型模糊模型,将其预测结果作为检验性能的基准。仿真结果表明,反向传播-相似度-奇异值分解混合迭代算法能够有效地消除冗余模糊集合与模糊规则带来的不良影响,基于其建立的预测模型具有较高的预测精度,能较好地跟踪实际负荷曲线,性能优于另外三个模型。According to the problem that it is difficult to precisely forecast power load with strongly random characteristic,the type-2 fuzzy logic method is introduced to reduce the prediction error.An interval type-2 fuzzy logic model was built to forecast short-term power load time series.The back-propagation（BP） arithmetic was used to tune model parameters;then,the similarity measure（SM） was applied to the simplification of redundant fuzzy sets;finally,the singular value decompose（SVD） method was proposed to choose optimal fuzzy rules.For test of the performance,a type-1 fuzzy logic model using the BP arithmetic,an interval type-2 fuzzy logic model using the BP arithmetic,and an interval type-2 fuzzy logic model using the SVD-BP hybrid iterative arithmetic were built as comparable benchmarks.The simulation results show that the SVD-SM-BP hybrid iterative arithmetic can effectively eliminate the harmful effects of redundant fuzzy sets and fuzzy rules,and the forecasting model based on it has better performance than other three models because it has higher prediction accuracy and better tracking characteristics for real power load curve. 更多

关 键 词：电力负荷预测 区间二型模糊逻辑 反向传播 相似度 奇异值分解 

分 类 号：TM715[电气工程—电力系统及自动化]