Study On Parameter Extraction Of Solar Cells Based On Numerical Optimization Algorithms

Solar photovoltaic technology is a new energy technology that uses photovoltaic modules to convert solar energy into electrical energy.It has received increasing attention in recent years and is considered an important means of achieving carbon peak and carbon neutrality goals.Research on the equivalent circuit model of solar cells and the extraction of related parameters is of great significance for the performance monitoring and automatic control of photovoltaic systems.However,existing parameter extraction nonlinear optimization algorithms still have problems such as reliance on initial value settings and high computational complexity.Therefore,how to build a more efficient and convenient solar cell parameter extraction method is still an open research topic.This paper is based on the single-diode model of solar cells.It first improves two existing nonlinear optimization algorithms,the Newton method and the particle swarm algorithm,by reducing the parameter dimension of initial value setting and implementing parameter reduction using information from three key data points: short circuit,open circuit,and maximum power point.This improves the fitting accuracy and ease of use of the original algorithms.Algorithm validation was performed using datasets from mainstream silicon and compound semiconductor solar cells,and results show that the improved algorithms have better performance and greater generality.For example,the root mean square error of the Newton method was reduced from2.4591E-03 to 1.7794E-03 in case 1,the number of iterations was reduced by 2,and the total computation time was reduced by 1 s.Secondly,a new linear parameter extraction algorithm is proposed based on the single-diode model.A linear relationship between voltage,current,and their relative derivatives is established,and a linear least squares method for solving the model parameters is introduced.This algorithm greatly reduces the computational complexity,does not require manually setting initial values or domain solving conditions,and can be fully automated.The effectiveness of the algorithm was also validated using the above-mentioned solar cell dataset and a subset of its data points,as well as a dataset of 600 outdoor photovoltaic components.In case 1,the root mean square error of this algorithm reached 7.8588E-04,which is better than the other nonlinear optimization algorithms and the above-mentioned improved algorithms.Thirdly,based on the second point,the algorithm is implemented in hardware using FPGA.Eight modules were designed based on the linear optimization algorithm proposed in this paper to implement a parameter extraction system that includes data import,parameter extraction calculation,and result output display,improving the practicality of the algorithm.Functional simulation and board-level verification showed that the root mean square error between the hardware and software calculation results differed by only 1.5%,and the system could accurately complete parameter extraction within tens of microseconds,making it possible to apply in edge computing scenarios for parameter extraction.