Research And Application Of Photovoltaic Array Failure Prediction Method

Photovoltaic array is an important part of photovoltaic power station,and its safe operation is not only the guarantee of normal power generation of photovoltaic power station,but also of great significance for reducing energy loss and troubleshooting potential safety hazards.With the application of intelligent technology in photovoltaic power generation industry,the research of related theories has laid a foundation for designing a fault prediction method suitable for photovoltaic arrays.This paper takes photovoltaic array as the research object.Firstly,the fault feature extraction,feature parameter prediction trend model and fault identification model of photovoltaic array are studied,and then they are applied to the fault prediction of photovoltaic array.This paper mainly includes the following work:First,the basic principle of solar power generation operation is studied,and the common faults and their causes of photovoltaic array are analyzed.By establishing the simulation model of photovoltaic array,the key fault characteristic parameters are extracted by comparing the output characteristic curves of normal state and fault state,and the influence of environmental factors on the characteristic parameters is analyzed.Second,the photovoltaic array experimental platform is built,and the design scheme,fault simulation method and data acquisition scheme of the experimental platform are described,and the collected data are classified to form a sample database.Third,the trend prediction algorithm of characteristic parameters and fault identification algorithm of photovoltaic array are studied.The parameters of support vector regression(SVR)algorithm are optimized by fruit fly optimization algorithm(FOA)optimized by three improved strategies,and the trend prediction model of photovoltaic array characteristic parameters based on MFOA-SVR is established.By optimizing the parameters of support vector machine(SVM)model through grid search and cross-validation,the fault identification model is established.Fourth,a combined model based on MFOA-SVR-SVM is proposed for fault prediction of photovoltaic array.Firstly,the collected data samples are input into the trend prediction model of photovoltaic array characteristic parameters in the simulation experiment,and the MFOA-SVR algorithm model with the best prediction performance is selected by comprehensively comparing the performance of the four prediction models.Then,the output vector of MFOA-SVR prediction model is input into SVM fault identification model with optimized parameters,and the state type is output to realize fault prediction.Finally,the feasibility and effectiveness of this method are verified by simulation experiments.