基于BSNN-ARX的光伏逆变器模型辨识  

Photovoltaic inverter model identification based on BSNN-ARX

作　　者：杨立滨[1] 张海宁[1] 李春来[1] 杨军[1] 王平[2] 

机构地区：[1]国网青海省电力公司电力科学研究院,青海西宁810008 [2]重庆大学输配电装备及系统安全与新技术国家重点实验室,重庆400044

出　　处：《现代电子技术》2017年第7期167-170,174,共5页Modern Electronics Technique

基　　金：青海省光伏发电并网技术重点实验室项目(2014-Z-Y34A)

摘　　要：光伏逆变器是光伏并网系统的核心部件,将基于Hammerstein模型的非线性系统辨识方法引入到光伏并网逆变器的建模中,把单相光伏并网逆变器视为双输入单输出的非线性黑箱系统。在Hammerstein模型的静态非线性环节采用B样条神经网络,动态线性环节采用ARX模型,同时采用基于误差学习准则和最小二乘递归准则的自适应学习方法。实验测试结果表明,提出的BSNN-ARX光伏逆变器模型辨识方法可以对不同天气条件下的逆变器输出功率进行高精度的辨识,从而为并网逆变器的建模提供一种有效途径。The photovoltaic inverter is the core component of the photovoltaic grid-connected system. A nonlinear system identification method based on Hammerstein model is introduced into the modeling of the photovoltaic grid-connected inverter, in which the single-phase photovohaic grid-connected inverter is considered as a nonlinear black-box system with dual inputs and single output. The B-spline neural network （BSNN） is adopted in the static nonlinear link of the Hammerstein model, and the ARX model is adopted in the dynamic linear link. The adaptive learning method based on error learning criterion and least square recursion criterion is employed. The experimental measuring results show that the photovoltaic inverter model identifica- tion method based on BSNN-ARX can identify the inverter＇s output power with high accuracy under different weather condi- tions, and provide an effective way to model the grid-connected inverter.

关 键 词：光伏逆变器 B样条神经网络 ARX模型 系统辨识 

分 类 号：TN711-34[电子电信—电路与系统] TM615[电气工程—电力系统及自动化]