Research On Hot Spot Recognition Method Of Photovoltaic Module Based On Generative Adversarial Networks

Solar energy,as a clean and renewable energy source,occupies an increasingly important position in the field of new energy applications.With the large-scale application of photovoltaic power generation technology,for the monitoring of working conditions in photovoltaic power generation processes,hidden dan gers of photovoltaic cells and the detection and monitoring of faults will certainly be an important link in the process of ensuring the safe production of photovoltaic power generation.The occurrence of hot spots in photovoltaic modules is a problem often encountered in the process of photovoltaic power generation.On the one hand,it will shorten the module’s power generation life,reduce the power generation output,and on the other hand,it will cause a fire,causing more serious economic losses.Therefore,it is of great significance to prevent and detect the hidden danger of hot spots in the production process of photovoltaic power generation for improving the reliability of power generation and reducing the cost.Based on the rapid development of image recognition technology and the application advantages of generating adversarial networks in the fiel d of data set enhancement and image classification,this paper proposes a hot spot identification and positioning method based on generating adversarial networks for the detection of hot spot of photovoltaic modules.The main contents of this method includ e:(1)pre-processing of the original photovoltaic infrared image;(2)production of photovoltaic cell infrared image data set;(3)multiple methods to enhance the cell image data set and compare the effects;(4)training generative adversarial networks and verifying classification effect;(5)designing a hot spot recognition and localization model based on generative adversarial networks.This method solves the problems of insufficient training and over-fitting caused by a small data set in an image recognition network.The recognition accuracy is high and the recognition process is more intelligent and efficient.In addition,the difference in temperature between modules is often used as a criterion for determining whether hot spots occur in photovoltaic production.Therefore,an evaluation standard based on the analysis of temperature is developed in this paper to evaluate the accuracy of the hot spot identification and positioning model.The experiment proves that the hot spot identification and positioning method proposed in this paper can accurately and efficiently i dentify and locate the hot spot area,the method has a good application value for the detection of hot spot faults in photovoltaic power stations and a reference value for the prevention of hot spot hazard.