Research On SOM Neural Network Algorithm For Online Fault Diagnosis Of Photovoltaic Array

With the emergence of energy shortages and environmental problems,the future energy structure of the world will inevitably tilt towards renewable energy,and the proportion of solar energy in renewable energy will reach a relatively high level.With the vigorous development of the photovoltaic industry,the proportion of the photovoltaic industry in power production will increase day by day.But various failures will inevitably occur during operation because the photovoltaic power stations are mostly built in complex environments,and effective diagnosis of photovoltaic array failures is a problem that needs to be solved urgently.The main research work of this paper is as follows:(1)The establishment of photovoltaic cell and photovoltaic array model.In this paper,the generation principle and mathematical model of photovoltaic cells are studied and analyzed.The simulation model of photovoltaic cells is established and the photovoltaic array model is further constructed in Matlab / Simulink,and the output characteristics of photovoltaic cells are analyzed.(2)The fault state of the photovoltaic array is simulated and the fault data is obtained.The four types of faults and faults causes of the photovoltaic arrays are studied.The four fault states of open circuit,short circuit,shadow and abnormal aging are simulated through the established photovoltaic array model.The output of the photovoltaic array in the fault state are obtained.Then the fault data are collected.By comparing with the output characteristics under normal conditions,the change parameters corresponding to each fault are judged,and then the input variables of the photovoltaic array fault diagnosis model are determined.(3)Studying the fault diagnosis of photovoltaic array.Taking the SP structure photovoltaic array as the research object,the photovoltaic array fault diagnosis model is established by using the self-organizing mapping neural network for fault diagnosis.The photovoltaic array fault data obtained by the simulation is normalized and input into the self-organizing mapping neural network for fault diagnosis,and then the four fault states of open circuit,short circuit,shadow and abnormal aging are detected.The simulation results show that the correct rate of the method to identify the fault state can reach more than 87%,the accuracy of diagnosis is higher than that of BP neural network and it can effectively complete the fault diagnosis of photovoltaic array.(4)An improved self-organizing map neural network based on dragonfly algorithm optimization is proposed.Aiming at the influence of the random selection of the initial weights of the self-organizing map neural network on the fault diagnosis results,a self-organizing map neural network optimized by the dragonfly algorithm is proposed.The dragonfly algorithm is used to optimize the initial weights of the self-organizing map neural network.The optimal weights obtained through optimization are used as the initial weights of the neural network,and the improved self-organizing map neural network is used to identify the failure mode of the photovoltaic array.The simulation result shows that,compared with the original self-organizing map neural network,the optimized neural network recognizes the fault status with a correct rate of 95.12%,which verifies the feasibility of the method.(5)The BP neural network is optimized by dragonfly algorithm,and the diagnosis results are compared with the self-organizing map neural network optimized by dragonfly algorithm to further verify the performance of the improved self-organizing map neural network.