利用多层感知机和Ⅰ-Ⅴ特性的光伏组件建模方法  

Photovoltaic module modeling method using multilayer perceptron and Ⅰ-Ⅴ characteristics

作　　者：余辉 陈志聪 郑巧 吴丽君 程树英 林培杰 YU Hui;CHEN Zhicong;ZHENG Qiao;WU Lijun;CHENG Shuying;LIN Peijie(Institute of Micro-Nano Devices and Solar Cells,College of Physics and Information Engineering,Fuzhou University,Fuzhou,Fujian 350108,China)

机构地区：[1]福州大学物理与信息工程学院,微纳器件与太阳能电池研究所,福建福州350108

出　　处：《福州大学学报（自然科学版）》2021年第3期336-342,共7页Journal of Fuzhou University(Natural Science Edition)

基　　金：国家自然科学基金资助项目(61601127);福建省科技厅高校产学合作资助项目(2016H6012);福建省科技厅引导性基金资助项目(2019H0006)。

摘　　要：为了提高光伏组件模型的准确度和可靠性,提出一种利用多层感知机和不同工况下实测的Ⅰ-Ⅴ特性曲线数据集的建模新方法.首先,使用双线性插值法对实测Ⅰ-Ⅴ曲线进行重采样,以提高Ⅰ-Ⅴ曲线上数据点分布的均匀性;然后使用基于温度-辐照度的网格采样法对数据集进行下采样,降低数据冗余度.其次,提出一种基于多层感知机神经网络的光伏组件模型,并基于预处理的Ⅰ-Ⅴ曲线数据集,使用Adam算法训练该模型.最后,采用实测Ⅰ-Ⅴ特性曲线数据集,验证和测试了所提出的建模方法,并与支持向量机、梯度提升决策树等机器学习算法进行对比.实验结果证明,所提出的建模方法具有更高的精度和泛化性能.In order to improve the accuracy and reliability of the photovoltaic module models, this paper proposes a new modeling method based on the multilayer perceptron and Ⅰ-Ⅴ characteristic curves dataset measured under different working conditions. Firstly, bilinear interpolation method is used for resampling the measured Ⅰ-Ⅴ curve to improve the uniformity of the distribution of data points on the Ⅰ-Ⅴ curve, and then the dataset was down-sampled by temperature-irradiance grid sampling method to reduce the data redundancy. Secondly, a new photovoltaic module model based on multilayer perceptron neural network is proposed, and based on the pre-processed Ⅰ-Ⅴ curve data set, the Adam algorithm is used to train the model. Finally, the measured Ⅰ-Ⅴ characteristic curve data set is used to verify and test the proposed modeling method, and compared with machine learning algorithms such as support vector machine and gradient boosting decision tree. Experimental results show that the proposed modeling method has the highest precision and generalization performance.

关 键 词：光伏建模 Ⅰ-Ⅴ特性 多层感知机 神经网络 

分 类 号：TM615[电气工程—电力系统及自动化]