Design And Implementation Of Solar Photovoltaic Array Identification And Hot Spot Detection Technology

With the rapid development of economy and continuous improvement of technology,fossil fuels not only fail to meet the growing energy demand,but also bring about serious environmental problems.Therefore,solar energy is regarded as one of the most important clean energy in the world.Solar power generation technology is mainly used in photovoltaic power plants,and hot spot effect is one of the major factors that affect the power generation efficiency of photovoltaic power plants.However,the efficiency and accuracy of traditional hot spot detection method are not high enough to meet the actual demand of the power plants.Therefore,the design of a high efficiency and accuracy hot spot detection method is of great significance.On the base of the infrared temperature data of power plant in operation,the photovoltaic array identification and hot spot detection are completed by this paper,including the following three points:(1)The improvement of photovoltaic array identification method based on adaptive threshold segmentation algorithm.Based on the prior knowledge of photovoltaic arrays with smaller temperature difference and backgrounds with greater temperature difference,the backgrounds whose temperature are low and temperature difference are greater,are pre-removed by reasonable setting of temperature threshold and local variance statistics.Based on this,the adaptive threshold segmentation method based on single Gaussian model is used to separate the photovoltaic arrays from the backgrounds,as the temperature histogram of photovoltaic arrays obeys Gaussian distribution.(2)The proposal of photovoltaic array identification method based on improved K-means algorithm.In order to deal effectively with the situation of small temperature difference between backgrounds and photovoltaic arrays,taking advantage of the strong correlation between adjacent frames and noise reduction by multi-frame cumulative,the photovoltaic arrays are separated from the backgrounds by the foreground modeling based on multi-frame infrared images.Furthermore,the traditional K-means algorithm is improved through the computation amount reduction by combining triangular inequality theorem and the optimization of initial clustering center choice by both sample ranking and density statistics of related samples.Then,According to the fact that the regional characteristics among the photovoltaic arrays are similar,the photovoltaic arrays are identified by the regional similarity measure based on the improved K-means algorithm.(3)The proposal of hot spot detection method based on SVM.In order to improve the efficiency of hot spot detection,the photovoltaic arrays are divided into several sub-image blocks by image segmentation technology.Then,the temperature features of sub-image blocks are extracted based on the infrared characteristics of hot spots and the hot spot detection is further implemented by SVM.In addition,the severity of hot spots is divided into three levels based on the temperature properties of the hot spots.The experimental results show that this paper not only realizes the effective identification of photovoltaic array,but also has higher detection efficiency and accuracy to satisfy the actual demand of photovoltaic power plants.Moreover,the classification of the severity of hot spots is conducive to further reduce the harm of hot spot effect.