Energy Conversion and Management ( IF 8.208 ) Pub Date : 2020-10-16 , DOI: 10.1016/j.enconman.2020.113521 Zhicong Chen; Yueda Lin; Lijun Wu; Shuying Cheng; Peijie Lin
In order to accurately and efficiently evaluate the actual operating status of photovoltaic (PV) power stations, this study proposes a novel design of quick current–voltage (I-V) curve tracer with automatic parameter extraction for PV arrays. The I-V tracer exploits the dynamic capacitor charging to quickly and safely scan the output I-V curve of PV array. Particularly, an adaptive sampling interval and charging and discharging time estimation method is proposed to achieve the uniform sampling of the I-V curve for improving the data quality. Moreover, an efficient grid search and improved Nelder-Mead simplex (GS-INMS) based hybrid optimization algorithm is proposed for PV model parameters identification. In the GS-INMS algorithm, the grid search method is used for finding starting point for the NMS optimization algorithm, which can improve the global search capability. Meanwhile, the basic NMS algorithm is modified to improve the local convergence. The performance of the proposed algorithm is verified and compared with some state-of-the-art methods on several experimental datasets, including two benchmark I-V curves, large dataset of I-V curves from the National Renewable Energy Laboratory (NREL), and the I-V curves collected by the designed I-V tracer system. Comprehensive comparison results demonstrate that the proposed GS-INMS shows obvious superiority in accuracy, convergence and robustness. Meanwhile, because of low complexity, this algorithm is embedded in the designed tracer for real-time parameters extraction of the single-diode model of PV arrays. In addition, the experimental results of field testing on a small-scale PV array indicate that the I-V tracer can quickly acquire I-V curves of high quality and perform accurate model parameter identification in real time.