摘要
针对短期风电功率预测,提出一种基于主成分分析(PCA)和一型非单值区间二型模糊逻辑系统(FLS)相结合的方法。PCA一型非单值区间二型FLS预测模型,应用反向传播(BP)算法设计预测模型前件和后件的参数,进一步将SVD-QR算法应用到BP方法应用于不同地区的风电场风电功率预测实例中,在同等条件下还分别与SVM(支持向量机)、一型非单值FLS、一型非单值区间二型FLS、PCA-单值区间二型FLS等其他预测方法进行比较。实验结果表明,所提方法取得了较高的预测精度,具有很好的预测效果,同时,模型的模糊规则数少,较好地解决了模糊模型的规则"爆炸"问题,这使得PCA-区间二型FLS方法在风电功率预测领域具有较好的应用潜力。
Aiming at short-term wind power forecasting,a method composed of principal component analysis(PCA)and interval type-2 fuzzy logic systems(FLS)with non-singleton type-1 fuzzification is proposed. PCA method is used to reduce input’s dimensions of the model. On the basis of this,taking into account the stochastic nature of wind power,a forecasting model using an interval type-2 FLS with non-singleton type-1 fuzzification is built. The back-propagation(BP)algorithm is used to update the parameters including input membership function, antecedent and consequent membership function respectively. Further more,SVD-QR algorithm is applied to the results of the BP algorithm to determine the reduced set of fuzzy rules,the process iterates until the forecast accuracy can meet design requirement.The employed method is then applied to real-world wind power forecasting instances in different areas. Under the same conditions,compared to the existing forecasting methods including type-1 FLS,interval type-2 FLS with singleton fuzzification,interval type-2 FLS with non-singleton type-1 fuzzification,PCA-interval type-2 FLS with singleton fuzzification,etc.,experiment results confirm that the employed method can achieve better forecasting accuracy.Meanwhile,fuzzy rule explosion problems are solved effectively owing to the reduced fuzzy set by PCA transform,hence,it shows good application potential in the wind power forecasting field.
作者
李军
王星辉
Li Jun;Wang Xinghui(School of Automation & Electrical Engineering, Lanzhou Jiaotong University, Lanzhou 730070, China)
出处
《太阳能学报》
EI
CAS
CSCD
北大核心
2019年第3期608-619,共12页
Acta Energiae Solaris Sinica
基金
国家自然科学基金(51467008)
关键词
风电功率
预测
主成分分析
区间二型模糊逻辑系统
BP算法
SVD-QR算法
wind power
forecasting
principal component analysis
interval type-2 fuzzy logic systems
back-propagation algorithm
SVD-QR algorithm