基于模糊集鲁棒概率不等式的风电斜坡估计方法  

WIND POWER SLOPE ESTIMATION METHOD BASED ON FUZZY SET ROBUST PROBABILITY INEQUALITY

作　　者：徐姣新 杨召 Xu Jiaoxin;Yang Zhao(Shangqiu University,Shangqiu 476000,Henan,China;Shangqiu Institute of Technology,Shangqiu 476000,Henan,China)

机构地区：[1]商丘学院,河南商丘476000 [2]商丘工学院,河南商丘476000

出　　处：《计算机应用与软件》2022年第3期108-115,共8页Computer Applications and Software

摘　　要：从不确定性量化的角度将风电出力的斜坡事件发生概率转化为一个数据驱动的鲁棒概率不等式,该不等式提供一个随机变量的概率,且其概率分布属于给定的多面体。为了解决风电输出的分布不确定性问题,考虑由可用数据构造的模糊集合中的候选分布集合,其最低要求是要包括预测值和平均绝对偏差。基于矩的模糊集由所有概率分布组成,共享相同的平均和平均绝对偏差值。当已知更多历史数据时,可以建立一个基于散度的模糊集,该模糊集包含所有接近Wasserstein度量意义上经验分布的概率分布。该方法提供最坏情况下的风电功率分布中斜坡事件的概率上限,历史数据越多,其保守性还可以显著降低。最后将所提方法与高斯混合模型进行比较,验证其有效性和优越性。From the perspective of uncertainty quantification,the probability of slope event of wind power output is transformed into a data-driven robust probability inequality,which provides the probability of a random variable and its probability distribution belongs to a given polyhedron.In order to solve the distribution uncertainty problem of wind power output,this paper considers the candidate distribution set in the fuzzy set constructed by available data.The minimum requirement is to include the predicted value and the average absolute deviation.The moment-based fuzzy set consists of all probability distribution,sharing the same mean and mean absolute deviation values.When more historical data is known,a fuzzy set based on divergence can be established,which contains all probability distributions close to the empirical distribution in the sense of Wasserstein metric.This method provides the upper limit of probability of slope events in the worst-case wind power distribution.The more historical data,the less conservative it is.Finally,the proposed method is compared with Gaussian mixture model to verify its effectiveness and superiority.

关 键 词：数据驱动优化 预测 斜坡事件 不确定性 风力发电 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]