基于相似日理论和CSO-WGPR的短期光伏发电功率预测  

Short-term Photovoltaic Power Generation Prediction Based on Similar Day Theory and CSO-WGPR

作　　者：孟安波[1] 陈嘉铭 黎湛联 丁伟锋 欧祖宏 殷豪[1] MENG Anbo;CHEN Jiaming;LI Zhanlian;DING Weifeng;OU Zuhong;YIN Hao(School of Automation,Guangdong University of Technology,Guangzhou 510006,China)

机构地区：[1]广东工业大学自动化学院,广州510006

出　　处：《高电压技术》2021年第4期1176-1184,共9页High Voltage Engineering

基　　金：国家自然科学基金(61876040)。

摘　　要：针对光伏发电功率预测精度不高的问题,提出一种结合纵横交叉算法与改进的高斯过程回归算法(crisscross optimization algorithm and weighted Gaussian process regression,CSO-WGPR)的预测模型。首先,通过加权模糊聚类对天气类型进行划分,选出与预测日相同类型的相似日样本;其次,采用单类支持向量机(One-Class supportvectormachine,One-ClassSVM)算法结合传统高斯过程回归算法,建立改进后的高斯过程回归模型(weighted Gaussianprocess regression,WGPR),减小异常值数据对预测结果的不良影响;然后,采用纵横交叉算法(crisscross optimization algorithm,CSO)优化WGPR的超参数,进一步提高模型的预测精度。以澳洲爱丽丝泉光伏系统为例进行建模预测,真实数据仿真和实验结果表明,所提预测模型在晴天、阴天、雨天类型下具有更高的预测精度,验证了该方法的有效性。Aiming at the problem of low accuracy of photovoltaic power generation prediction, a prediction model combining crisscross optimization algorithm and weighted Gaussian process regression algorithm(CSO-WGPR) is proposed. Firstly, the weather types are divided by weighted fuzzy clustering, and similar day samples of the same type as the forecast days are selected. Secondly, one-class support vector machine(One-Class SVM) algorithm combined with traditional gaussian process regression algorithm is used to establish a weighted Gaussian process regression model(WGPR) to reduce the adverse effects of outlier data on the prediction results. Finally, the crisscross optimization algorithm(CSO) is used to optimize the hyperparameters of WGPR to further improve the prediction accuracy of the model. Australian Alice Springs photovoltaic system is taken as an example for modeling and prediction, and the real data simulation and experimental results show that the proposed prediction model has higher prediction accuracy under sunny, cloudy, and rainy days, which verifies the effectiveness of the method.

关 键 词：光伏发电 功率预测 加权模糊聚类 单类支持向量机 改进的高斯过程回归 纵横交叉算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TM615[自动化与计算机技术—控制科学与工程]