The parameters extraction method of the single diode model is usually used to analyse silicon-based solar cells to predict their behaviour from IV curves. However, the film solar cells based on compound semiconductor, for example the CdTe and CIGS, which have been widely used in the market, have few relevant studies still now. This paper presents two parameters extraction methods and applies to two photovoltaic modules, then we use a function in MATLAB to refine the results. Finally, the error of the two methods has been compared to find the best fitting result. Final fitting results show that these methods have their own advantages and disadvantages and both have great fitting accuracy on the two modules, and they can be used in various irradiance conditions.
The IV characteristics of photovoltaic (PV) modules are an important factor during the process of production and use. In recent years, many related fitting studies have been done. It is a typical regression analysis problem and the main research method is curve fitting. Research objects are usually the silicon-based solar cells, and the average errors are usually between 10-3 and 10-4. But it is seldom used in CdTe and CIGS solar cells.
Considering the problem of local optimization, as a common global optimization algorithm, PSO can be used to solve it. We can assume the problem as a point with coordinates n and Rs, looking for the deepest pit on the plane with n and Rs, in which the deepest pit means the smallest SE.
This method can let different number of points search the deep pit along with different directions, in different velocities. Recording the depth every time to get the deepest pit which means the global optimal solution with compare. The logic chart is shown in figure 2.
We choose two kinds of PV modules’ data, which is from National Renewable Energy Laboratory, Golden, CO, USA, to test the two methods. They are two practical film solar cells, and one is based on CdTe, the other one is based on CIGS. In order to get the more accurate results, a refinement step has been added after that. The optimization function "lsqnonlin" in MATLAB is good for the least square problem, and the previous results can be substituted in it as initial values to get refined results.
Fig2
We choose two kinds of PV modules’ data, which is from National Renewable Energy Laboratory, Golden, CO, USA , to test the two methods. They are two practical film solar cells, and one is based on CdTe, the other one is based on CIGS. In order to get the more accurate results, a refinement step has been added after that. The optimization function "lsqnonlin" in MATLAB is good for the least square problem, and the previous results can be substituted in it as initial values to get refined results.
The fitting results and error analysis show that: (1) Five parameters of compound semiconductor PV module can be extracted out accurately by the three methods; (2) No negative values appear, which means that results are in accords with their physical meanings, SDM is also suitable for CdTe and CIGS solar cells; (3) NR has faster solution, but the accuracy of PSO is better than NR; (4) CIGS module has worse fitting result than CdTe module, maybe other algorithm can solve this problem.
上一篇： 元素和化合物半导体薄膜的激光退火
