Research And Implementation Of Photovoltaic Array Fault Detection Based On Convolutional Neural Network

Solar energy is a typical renewable energy source and one of the new energy sources with the most development potential today.It has the characteristics of large total amount,easy access,and unexploited application.As the energy revolution continues to deepen in my country and globally,the new energy industry represented by photovoltaic power generation is gradually moving towards a more mature stage of development.In recent years,photovoltaic power generation has gained sustained development and attention on the world stage,and vigorously promoting the development of the photovoltaic industry is the top priority of today’s energy policy.As one of the core components of the photovoltaic system,the normal operation of the photovoltaic array has a decisive effect on the photovoltaic system.The development of the photovoltaic array fault detection model is urgent.In summary,this article mainly completed the following work around photovoltaic array fault detection:First,start with the energy structure and environmental issues,and combine the development status of the photovoltaic industry and fault detection at home and abroad to establish the feasibility of this subject research.Secondly,the discussion is based on the type and structure of photovoltaic arrays.Through the introduction of photovoltaic system power generation principles and equivalent circuits,the manifestations,generation principles and mechanisms of photovoltaic array failures are analyzed,and the photovoltaic array failure simulation model is established.The operating characteristics of the equipment are analyzed and the imbalance of the ratio of normal data to fault data is analyzed,combined with the above conditions,environmental data in different intervals are set to obtain photovoltaic array operating data,and the actual operating data of a photovoltaic base is compared to verify it.Availability of simulation models.Then,the method and principle of the photovoltaic array fault detection model are introduced.Support Vector Machine(SVM)and K-Nearst Neighbors(K-Nearst Neighbors,KNN),which are suitable for the research content of this subject,are used to search for them.The detection results are output after optimization and tuning;because the Synthetic Minority Oversampling Technique(SMOTE)algorithm is prone to sample stacking,it fails to provide useful information and the limitations of the sample.After the algorithm is improved,the data will be unbalanced.Combining the advantages of convolutional neural network with local connection and weight sharing,a fault detection model that optimizes convolutional neural network through grid search method nested cross-validation method is proposed,and the experimental results of other models are compared,The rationality and effectiveness of the model in terms of accuracy and time performance are verified.Finally,using the photovoltaic array fault detection model proposed in this paper,the photovoltaic array fault detection system is designed to realize the query function of the photovoltaic array operating data and the photovoltaic array fault detection function.