Multi-model Parameters Identification And Modeling Of Photovoltaic Cells Based On Neural Network

In recent years,with global economic development and scientific and technological progress,the demand for electricity has steadily increased year by year.The use of non-renewable energy for power generation has not only led to the gradual exhaustion of non-renewable energy,but also aggravated environmental pollution.Since renewable energy is inexhaustible and does not pollute the environment,how to efficiently use renewable clean energy for power generation has attracted widespread attention from countries around the world.As one of the main renewable energy sources,solar power,photovoltaic power generation has attracted a lot of attention and wide application around the world,but the understanding of the relationship between the physical model parameters of photovoltaic cells and the irradiance are not uniform,and there are not many studies on the shape model parameters and irradiance,temperature.Therefore,how to establish an effective and accurate photovoltaic cell equivalent model,how to obtain the relationship between photovoltaic cell model parameters and irradiance,and how to obtain the photovoltaic output characteristic curves under a certain condition has attracted widespread attention.Maximum power points tracking have an important impact on the reliability of the power grid.This paper introduces the main principles of photovoltaic cell power generation,the advantages and disadvantages of the physical and mathematical models of photovoltaic cells.Each parameter in the physical model has its own physical meaning,but the expression is an implicit equation,the solution process is complicated,and the time is relatively long.The mathematical model is established based on the shape of the?-? curve,each parameter has no clear physical meaning,and the equation solving is simple and fast.By comparing the accuracy of different solution methods,the particle swarm algorithm with the highest solution parameter accuracy is selected as the parameter identification method in this paper.This paper uses data to verify and analyze the relationship between the physical model parameters and the irradiance,which are currently difficult to form a unified answer,and also conduct a detailed analysis and discussion on the specific relationship between the shape model parameters and the irradiance that are not studied much at present.Secondly,this paper mainly uses the power exponential shape model,and proposes a method to establish a photovoltaic cell model based on the illumination classification and the training data range is larger than the calculation range of the neural network.This paper focuses on the above discussion in two aspects,unified analysis and explanation of the relationship between the photovoltaic cell model and the irradiance,and provides new ideas for the method of building photovoltaic models using neural networks:(1)Introduce the mapping function in the BP neural network,how to select the number of hidden layers,use the known irradiance and the model parameters of the photovoltaic cell to train the neural network,use error function to calculate the accuracy of different methods and perform analysis to prove the validity and reliability of the established model.Mainly use BP neural network to model the two commonly used physical models and two shape models,find the relationship between the model and the light temperature for analysis.Unified consideration of the relationship between the current physical model parameters and irradiance,to make up for the current gap in the research on the relationship between shape parameters and irradiance.(2)On the basis of Chapter 3,verify the method proposed in Chapter 4 based on the shape model.Firstly,in the process of establishing the neural network,the irradiance of the training data is divided into two parts of high irradiance and low irradiance,and the method of overlapping the boundaries of the two neural networks is used to expand the training range of the neural network.Using the trained neural network,different neural networks are used to establish the output characteristic curves of the photovoltaic cell according to the different irradiance conditions.Finally,the output characteristic curves measured by the four different types of photovoltaic cells under actual conditions are compared with the output characteristic curves calculated by the method proposed in this chapter,and it is concluded that the method proposed in this chapter improves the establishment of photovoltaic cell models and output characteristic curves accuracy.