Estimação de Parâmetros de Modelo de Sistemas Fotovoltaicos utilizando Algoritmo de Levenberg-Marquardt Modificado

ABSTRACT This paper proposes an algorithm for the estimation of the parameters of a photovoltaic system model. We propose to use a modified Levenberg-Marquardt algorithm to enhance the estimates of the two resistences (series and parallel) in an one-diode photovoltaic model. The estimation precision of the proposed approach is compared to other three methods. The simulation of the photovoltaic systems using the obtained parameters provides the characteristic curve and the power generated by the photovoltaic system. Methods are applied to interpolate the points of the IxV curves obtained by simulation, so that these IxV curve can be compared with the IxV curves provided by the datasheet of the Kyocera KC200GT photovoltaic module. The results show that the curve generated with the parameters estimated through the proposed algorithm presents the smallest mean square error compared to the other considered methods.

摘要 本研究提出一种用于光伏系统模型参数估计的算法。我们提出采用改进的莱文贝格-马夸尔特（Levenberg-Marquardt）算法，以优化单二极管光伏模型中串联电阻与并联电阻的估计结果。将所提方法的估计精度与其余三种方法进行对比。利用所得参数对光伏系统进行仿真，可得到光伏系统的特性曲线与输出功率。本文采用前述四种方法对仿真得到的电流-电压（IxV）曲线数据点进行插值，以便将这些曲线与京瓷（Kyocera）KC200GT光伏组件数据手册中提供的电流-电压（IxV）曲线进行对比。结果表明，相较于其余所考量的方法，基于本研究算法估计参数生成的曲线具有最小的均方误差。