基于相似时刻的光伏出力概率分布估计方法

发布时间：2018-09-14 14:47

【摘要】：针对光伏发电可预测性低的问题,提出了一种综合使用通径分析(path analysis,PA)、k近邻算法(k-Nearest Neighbor,KNN)、神经网络分位数回归(quantile regression neural network,QRNN)和核密度估计(kernel density estimator,KDE)的光伏出力概率分布估计方法,构造出未来1 d任意时刻的光伏出力概率密度函数,可以得到比点预测和区间预测更多的有用信息。首先由通径分析对气象因素进行约减,在降低模型输入维数的基础上减小变量间的耦合作用。然后通过K-means算法按天气类型对历史样本进行聚类,进一步提高相似样本的筛选效果。最后利用神经网络分位数回归和核密度估计对光伏出力的概率分布进行估计。实验结果表明,相比于核密度估计和传统的正态分布估计方法,采用所提方法估计出的概率分布的可靠性和锐度更高。
[Abstract]:Aiming at the problem of low predictability of photovoltaic generation, this paper presents a method for estimating the probability distribution of photovoltaic force using path analysis (path analysis,PA) algorithm (k-Nearest Neighbor,KNN), neural network quantile regression (quantile regression neural network,QRNN) and kernel density estimation (kernel density estimator,KDE). The probabilistic density function of photovoltaic force at any time in the next 1 day can obtain more useful information than point prediction and interval prediction. Firstly, the meteorological factors are reduced by path analysis, and the coupling between variables is reduced on the basis of reducing the input dimension of the model. Then the K-means algorithm is used to cluster the historical samples according to the weather type to further improve the screening effect of similar samples. Finally, the probability distribution of photovoltaic force is estimated by neural network quantile regression and kernel density estimation. The experimental results show that the reliability and sharpness of the probability distribution estimated by the proposed method are higher than that of the kernel density estimation method and the traditional normal distribution estimation method.
【作者单位】：天津大学电气工程与自动化学院;天津内燃机研究所;
【基金】：国家自然科学基金项目(61374122)~~
【分类号】：TM615

【相似文献】