Research On Photovoltaic Array Fault Diagnosis Method Based On Integrated Learning

With the pollution and destruction of traditional energy to the environment becoming more and more prominent in the past,and the decline of fossil fuels,solar energy has become an important part of human energy use,and has been continuously developed.Driven by the third industrial revolution,China has launched a large number of policies on clean energy such as wind energy and solar energy to encourage and develop the new energy industry.Solar energy with the advantages of renewable,clean and environmental protection has quickly become a representative of new energy.Photovoltaic modules are an indispensable part of photovoltaic power generation system.To ensure the safe and stable operation of photovoltaic modules and timely fault alarm are the prerequisite to improve the power generation capacity and prevent major accidents such as accidental fire.Therefore,this paper focuses on photovoltaic modules,analyzes possible faults of photovoltaic modules in the operation of photovoltaic modules,and selects PNN as the basic fault diagnosis model based on the excellent pattern recognition capability of probabilistic neural network(PNN).Sines and cosines optimization(SCA)algorithm is used to optimize the smoothing factor of PNN,so as to solve the accuracy deviation caused by setting parameters artificially.Based on the AdaBoost integrated algorithm model,the SCA-optimized PNN is used as the weak classifier,and the output results of the strong classifier are obtained by using the probabilistic linear combination of various weak classifiers,so as to improve the accuracy of model diagnosis.The main contents of this paper are as follows:(1)By analyzing the mathematical model of photovoltaic cells and establishing the corresponding circuit simulation model on MATLAB platform,the output characteristics of photovoltaic modules under the condition of uneven illumination and different temperatures and under normal conditions are analyzed through the model,and the trend of its change characteristics is pointed out.(2)The common types of main faults(open circuit,abnormal aging,blocking shadow and short circuit)of photovoltaic modules are summarized,and the causes and performance characteristics of faults are summarized.A 3x3 photovoltaic array model is built,and different fault states are simulated to obtain the output characteristic curve of photovoltaic array under different fault conditions.The main parameters affecting the photovoltaic array are analyzed and enumerated,and then the fault diagnosis characteristic parameters are used as diagnosis parameters.(3)In view of the limitations of the traditional probabilistic neural network that needs to be set artificially and constantly try smoothing factors,SCA algorithm is used to optimize PNN smoothing factors to form an adaptive probabilistic neural network(SPNN).On the basis of this model,AdaBoost is introduced to take SPNN as weak classifier,and the classification results are obtained by linear combination of output probability.Compared with the traditional BP algorithm,the diagnosis accuracy of ASPNN model is greatly improved.The effectiveness and efficiency of the improved model are verified.