2参数辨识算法 - 福州大学学报（自然科学版.docVIP

改进多种群粒子群算法辨识光伏组件参数 吴越，，, 林培杰，程树英，陆培民 （福州大学微纳器件与太阳能电池研究所，福建 福州 350116） 针对光伏组件参数辨识问题，本文首先通过调整光伏单二极管重构目标函数，又预估计模型参数对有哪些信誉好的足球投注网站空间进行优化，再结合多种群粒子群算法与单纯算法的优点，构造出N-MPSO算法用于光伏组件模型参数的辨识。最后利用多种实际光伏组件测量数据对所提方法进行检验。结果表明N-MPSO算法相较于传统算法能够更加准确快速且稳定辨识出任意环境条件下光伏组件的模型参数对于光伏组件的设计具有实际意义。 关键词：光伏组件；参数辨识；N-MPSO算法 TP391 Photovoltaic module parameters identification using an improved multi-group particle swarm optimization algorithm Yue Wu, Zhicong Chen, Lijun Wu, Peijie Lin, Shuying Cheng, Peimin Lu （Institute of Micro-Nano Devices Solar Cells, Fuzhou University, Fuzhou, 350116, China） Abstract： the issue of photovoltaic module parameters identification, a new hybrid algorithm based on multi-group particle swarm optimization and simplex method is proposed. Firstly, the transcendental equation of the single diode photovoltaic model is modified so as to greatly reduce the computation complexity. Secondly, the search space for the parameters is optimized by pre-estimating the parameters initial value. And then, combining the advantage of multi-group particle swarm optimization and simplex method, a hybrid N-MPSO algorithm is constructed to quickly obtain the stable and accurate parameters. Finally, the algorithm is validated by several groups of I-V data measured from some typical photovoltaic modules. The results show that the proposed N-MPSO algorithm can reach a higher accuracy and lower time complexity compared with some other conventional methods, which is significant to the design, testing and diagnosis of photovoltaic modules and power stations. Key words：PV module；Parameter identification；N-MPSO 0引言 光伏电池及组件的物理特性数学模型是分析和设计光伏发电系统的重要基础。其中单二极管光伏模型能较准确描述实际光伏电池组件的非线性输出特性，因而在实际工程中得到了广泛的应用。但该模型属于隐式超越方程，其I-V特性无法直接用显式表达式描述，为其模型参数的提取带来难度[1,2]。 由于智能优化算法不断地发展与改进，近年来越来越多被应用到光伏组件模型的参数辨识，如模拟退火算法(SA)[3]、粒子群算法PSO）[4]、模式有哪些信誉好的足球投注网站算法PS）[5]、混沌无性繁殖优化算法CARO）[6]，这些算法在参数辨识中的效果显著，对比数学分析参数近似求解的方法在精度上有明显的优势。但其中SA、PSO算法的收敛速度慢，易过早陷入局部极值的问题，影响其求解的精度和可靠性。为了更加精确对模型参数求解，文献[7]在人工智能算法的基础之上，引入LambertW函数构造出显示单二极管模型作为算法的适应度函数，大幅提高了拟合精度。但