Research On Intelligent Detection Method Of Photovoltaic Array Shadow Occlusion Based On Support Vector Machine

In the process of photovoltaic power generation,the phenomenon of array being obscured by shadow is inevitable,which will lead to the output power of photovoltaic cells mismatch,serious will appear "hot spots",so that the efficiency of photovoltaic power generation reduced at the same time for the power station has brought security risks.The study of the output characteristics of the array under different degrees of shadow masking can optimize the maximum power extreme point tracking algorithm under shadow masking and improve the output efficiency of the power generation system,on the other hand,it is of practical significance to detect whether the array itself is in the shadow masking environment in real time according to the output characteristic curve,which is of practical significance for reducing the power loss of the photovoltaic system and maintaining the safe operation of the power station.The main contents of the thesis include:(1)In view of the output power factors affecting the photovoltaic array——solar irradiance and ambient temperature,the array model was constructed using Simulink simulation platform,and the out-of-output characteristics of the array under the action of the two were analyzed and compared.(2)Aiming at the problems such as the traditional maximum power point tracking method under shadow masking,which is prone to oscillation and low search accuracy at runtime,a rolling output voltage strategy and reverse voltage-seeking strategy are proposed,and the variable step perturbation observation method is combined to improve the problem of obvious shock in the output of the array characteristics.(3)In the light of the problems of low accuracy and prone to overfitting when detecting array shadow occlusion by traditional neural network methods,based on the improved maximum power extreme point tracking algorithm to obtain the data set,a Linear Discriminant Analysis-Artificial Bee Colony-Support Vector Machine(LDA-ABC-SVM)shadow occlusion detection method with high accuracy is proposed.After simulation verification,(1)Under the two conditions of fixed irradiance and mutative irradiance,using the improved maximum power extreme point tracking algorithm in the array both have an admirable tracking accuracy,effectively avoid the traditional closed-loop perturbing tracking method of the existing concussion problems.(2)When LDA-ABC-SVM method is used to detect different degrees of shadow masking,it can not merely reduce the dimension of the input data of high dimensions,but also make the time complexity of the detection algorithm reduced,the detection accuracy is improved when the array is in different degrees of shadow occlusion,and the problems of long training time is effectively solved when the commonly used neural network method is used in detection.