Research On Photovoltaic System Parameter Identification Based On Intelligent Optimization Algorithm

Energy is an important driving force for the development and progress of today’s society.Fossil energy has been developed and used in large quantities,but fossil energy is a non-renewable energy source and brings greenhouse effect and environmental pollution,so renewable energy sources are used more and more widely.Solar energy is an important renewable energy source,and one of the best current solar energy utilization devices is photovoltaic cells.It has been very popular to add the battery into the photovoltaic system as an auxiliary power source for the photovoltaic cell operation.In order to optimize the output power of the photovoltaic cell to track the maximum power point,and monitor the state of the battery when it is operating,it is especially important to identify the internal parameters of the cell model in the photovoltaic system.In this thesis,the intelligent optimization algorithm is studied to solve the problem of identifying the model parameters of photovoltaic cells and batteries in the photovoltaic system,so as to improve the accuracy of parameter identification.The main research contents are as follows:(1)When using intelligent optimization algorithms to solve the problem of model parameter identification in the photovoltaic system,the dynamic opposite learning strategy is designed to improve the problem in order to avoid the situation where the results converge to a local optimum.The search area of each generation of particles is adjusted dynamically,the quality of results is refined,and the probability of finding the global optimal solution for the particles in the population is increased.(2)In the photovoltaic system,the dynamic opposite learning strategy is applied to identify the unknown parameters of the models.For the elite particle population and the obsolete particle population in the algorithm,the dynamic opposite learning strategy is adopted to improve the problem of insufficient individual capabilities of different populations.The opposite learning method is used to improve the particle’s exploitation ability at the early stage of search.In the later stage,and the search space of particles is dynamically adjusted according to the learning ability of particles at the later stage to improve the exploration ability of particles.(3)The photovoltaic cell’s equivalent circuit models are analyzed to determine the internal parameters and objective functions to be extracted.The experimental data of different types of photovoltaic cells are combined.The improved algorithm is used to perform parameter identification experiments in the photovoltaic cell models.The simulation experiment of maximum power point tracking is achieved by using the identified model parameters.The parameter identification results and capabilities of different algorithms are compared.The result distributions when different algorithms identify the model parameters are analyzed.The effects of different populations adopting dynamic opposite learning strategy are tested.The results show that the improved intelligent optimization algorithm has more accurate results in identifying model parameters of the photovoltaic cells,and the efficiency of the algorithm is better than other algorithms,which can solve the problem of identifying model parameters of the photovoltaic cells.(4)The different battery models are compared,and the battery model for the parameter identification experiment is determined.The battery output equation is analyzed,and the objective function of parameter identification is obtained.The experimental data are collected from the discharge experiment,and used for parameter identification experiment.The output voltage curve and voltage error curve of the battery are constructed according to the parameters of the identification model.The identification results and performance of the improved algorithm are compared with the similar algorithms,and the effects of dynamic opposite learning strategy in different populations are analyzed.The results show that under the premise of ensuring its own performance,the improved algorithm can obtain better battery model parameters.The error between the calculated output voltage and the actual voltage is small,which can satisfy the requirements of the system to monitor the working condition of the battery.