Fault Detection And Diagnosis Of Photovoltaic Modules Based On Model And Data Driven

With the rapid development of economy and society and the increasing population of the world,the demand and consumption of energy are becoming larger and larger.The shortage of fossil energy and the harsh environment and climate make human eyes turn to the exploration and development of renewable energy.As one of the most important renewable energy sources,solar energy has been widely used in electric power production.At the same time,the failure of photovoltaic module,an important part of photovoltaic system,is becoming more and more prominent in the operation process.This paper takes the photovoltaic experiment platform as the research object,takes the simulation model and experimental data as the analysis method,adopts the data drive and data mining technology,and conducts the following research on the fault impact and fault diagnosis of photovoltaic modules:Firstly,a large number of operation data of modules and environmental data in photovoltaic experimental platform were collected,and the influence of light intensity and ambient temperature on the power generation of photovoltaic modules was analyzed.It was proved that both of them are positively correlated with power generation.Local shading experiments of different types and areas were carried out on photovoltaic modules to obtain the variation of output characteristics and power loss of the modules with the shielding areaSecondly,a simulation model of photovoltaic experiment platform based on Matlab/Simulink was established.Based on the analysis of the mathematical model of photovoltaic cells,the four-parameter method was used to build and validate the simulation model of photovoltaic modules in the photovoltaic experimental platform.Based on the maximum power point tracking technology and the principle of micro inverter,the variable step perturbation observation method and flyback converter were used to build the simulation model of micro inverter,and the simulation was verified.Finally,the fault simulation experiments of short circuit,open circuit,aging and partial shading were carried out on the photovoltaic experimental platform and simulation model,and the fault data samples of the experiment and simulation were collected.Because of the simple structure,high accuracy,fast learning speed and good generalization performance of Extreme Learning Machine(ELM),a kernel ELM classification model based on Improved Particle Swarm Optimization(IPSO)algorithm was proposed to classify the fault data.And four different optimization algorithms were used to verify the effectiveness of the fault diagnosis model.