In the scope of solar energy-based electrical needs in rural tropical regions, the present article develops and confronts experimental power models from the using of manufacturer data and a new model made with the met...In the scope of solar energy-based electrical needs in rural tropical regions, the present article develops and confronts experimental power models from the using of manufacturer data and a new model made with the meteorological and electrical data acquired. These data are registered through an acquisition station around a monocrystalline photovoltaic panel, designed and realized in the scope of this work. After the acquisition of meteorological data, a choice of the most relevant meteorological variable as input vectors to express the output powers obtained was carried out. Around the Single-Diode model, seven models are performed with analytics equations, iterative methods and an optimization method with a multi-objective function to get internal parameters. The proposed experimental model is made by a combination of the solution got at STC of an iterative method, with the value of nameplate and the use of an open circuit voltage equation with experimental coefficient to predict power output in operating conditions, and it’s demonstrated more efficient. The optimization of a multi-objective function using Nonlinear Squares (NLS) through the Leveng-Marqued method to solve the parameter estimation of a PV panel has been well done and the results are useful, like classic iterative method and less time-consuming.展开更多
随着“双碳”目标的推进,清洁能源所占比重大幅度增加,分布式光伏发电在我国农村地区快速发展,但其随机性、间歇性的特点给新能源消纳和电网稳定带来很大的挑战。光伏发电预测可以在一定程度上改善新能源消纳问题,减少光伏发电的不稳定...随着“双碳”目标的推进,清洁能源所占比重大幅度增加,分布式光伏发电在我国农村地区快速发展,但其随机性、间歇性的特点给新能源消纳和电网稳定带来很大的挑战。光伏发电预测可以在一定程度上改善新能源消纳问题,减少光伏发电的不稳定性对电网的冲击。因此,为提高光伏发电功率预测精度,提出一种基于改进向量加权平均算法优化CNN-QRGRU网络的光伏发电概率预测方法。首先采用ReliefF算法对特征变量进行选择,在此基础上利用高斯混合模型(Gaussian mixture model,GMM)聚类方法将天气分为晴天、晴转多云和阴雨天3种类型,将处理好的数据输入到CNN-GRU模型中,并利用向量加权平均(weighted mean of vectors algorithm,INFO)优化算法对模型超参数进行调参,将分位数回归模型(quantile regression,QR)与INFO-CNN-GRU模型相结合得到光伏功率条件分布,结合核密度估计法从条件分布中获得概率密度函数,完成概率预测。以实际光伏电站数据作为基础,将提出的INFO优化算法与其他几种传统的优化算法进行对比,结果表明INFO的优化效果更好,在此基础上进行概率预测,得到的概率预测结果相较于点预测能提供更多有效信息,更具有应用价值。展开更多
文摘In the scope of solar energy-based electrical needs in rural tropical regions, the present article develops and confronts experimental power models from the using of manufacturer data and a new model made with the meteorological and electrical data acquired. These data are registered through an acquisition station around a monocrystalline photovoltaic panel, designed and realized in the scope of this work. After the acquisition of meteorological data, a choice of the most relevant meteorological variable as input vectors to express the output powers obtained was carried out. Around the Single-Diode model, seven models are performed with analytics equations, iterative methods and an optimization method with a multi-objective function to get internal parameters. The proposed experimental model is made by a combination of the solution got at STC of an iterative method, with the value of nameplate and the use of an open circuit voltage equation with experimental coefficient to predict power output in operating conditions, and it’s demonstrated more efficient. The optimization of a multi-objective function using Nonlinear Squares (NLS) through the Leveng-Marqued method to solve the parameter estimation of a PV panel has been well done and the results are useful, like classic iterative method and less time-consuming.
文摘随着“双碳”目标的推进,清洁能源所占比重大幅度增加,分布式光伏发电在我国农村地区快速发展,但其随机性、间歇性的特点给新能源消纳和电网稳定带来很大的挑战。光伏发电预测可以在一定程度上改善新能源消纳问题,减少光伏发电的不稳定性对电网的冲击。因此,为提高光伏发电功率预测精度,提出一种基于改进向量加权平均算法优化CNN-QRGRU网络的光伏发电概率预测方法。首先采用ReliefF算法对特征变量进行选择,在此基础上利用高斯混合模型(Gaussian mixture model,GMM)聚类方法将天气分为晴天、晴转多云和阴雨天3种类型,将处理好的数据输入到CNN-GRU模型中,并利用向量加权平均(weighted mean of vectors algorithm,INFO)优化算法对模型超参数进行调参,将分位数回归模型(quantile regression,QR)与INFO-CNN-GRU模型相结合得到光伏功率条件分布,结合核密度估计法从条件分布中获得概率密度函数,完成概率预测。以实际光伏电站数据作为基础,将提出的INFO优化算法与其他几种传统的优化算法进行对比,结果表明INFO的优化效果更好,在此基础上进行概率预测,得到的概率预测结果相较于点预测能提供更多有效信息,更具有应用价值。