Research On Photovoltaic Module Fault Detection Based On Data Driven

With the continuous consumption of fossil energy such as coal and natural gas and the increasing environmental problems,new energy power generation has become the main form of energy substitution and power supply in the world.As a clean and environment-friendly energy,solar energy has developed rapidly in recent years.The installed capacity of photovoltaic power stations in China is increasing year by year,reaching 204.3gw by the end of 2019.The northwest region with excellent solar energy resource endowment is preferred for the construction of photovoltaic power station,but the population density of this kind of region is low,and the harsh environmental conditions such as desert,plateau and Gobi will have an adverse impact on the photovoltaic module,which greatly increases the probability of failure.In addition,the operation and maintenance is difficult,and the operation and maintenance cycle is long,which seriously affects the power generation of photovoltaic power station.The operation life of photovoltaic modules is generally25-30 years,but due to the bad environment and other reasons,the life of modules is greatly reduced.According to statistics,the annual average power attenuation in different areas and different environments is between 0.6% and 1%,which greatly affects the power generation efficiency level of the whole photovoltaic power station.Therefore,in order to ensure the maximum efficiency output of photovoltaic power station,it is a great challenge to photovoltaic operation and maintenance.Module fault detection method has become the key technology of photovoltaic power station operation and maintenance,and it is the key research direction to help photovoltaic power station fine operation and maintenance management and improve system energy efficiency.Firstly,according to the research direction of photovoltaic module fault detection,the detection methods are divided into five categories: Based on circuit structure,based on I-V curve scanning,based on infrared image detection,based on mathematical model and based on intelligent algorithm.Then,based on the principle of photovoltaic power generation,the module of photovoltaic power generation is simulated by Matlab / Simulink,and the module model is used to simulate the photovoltaic array,and the I-V output characteristic curve of photovoltaic array is obtained.According to the output characteristics of photovoltaic array,the maximum power tracking of photovoltaic array under shadow condition is studied,and the formula of photovoltaic array under shadow condition is analyzed.According to the output characteristics of photovoltaic array under multi peak condition,ant colony algorithm is used to track the maximum power of photovoltaic array under shadow condition.Secondly,the typical faults are simulated,and the changes of corresponding electrical parameters under each fault condition are analyzed,so as to obtain the fault characteristic parameters.The fault simulation data are input by the fault simulation model for fault detection.Because the fault data is sensitive to the first principal component,based on this characteristic,the principal component analysis method is used to preliminarily identify the normal data and fault data.Due to the large order of data,the influence of single fault data is small,so the oversampling principal component analysis method is used to improve the identification ability of fault data.Secondly,the fault data is detected twice,and the fuzzy clustering algorithm is used for fault detection and classification.Because the fuzzy clustering algorithm is easy to fall into local minimum,the adaptive chaos particle swarm optimization method is used to optimize the fuzzy clustering detection method,which greatly improves the fault detection accuracy.Finally,the fault simulation experiment is carried out by using the experimental platform.After obtaining the experimental fault data,the fault detection method is verified.After the identification of this method,it is proved that this method has good fault recognition effect on several types of photovoltaic array faults.