Research And Implementation Of A Photovoltaic Module Hot Spot Detection System Based On Infrared Image Recognition

The photovoltaic thermal spot failure is a phenomenon that some part of cells become a load to consume the energy of other photovoltaic components and generate heat due to changes in cells’ own features during the operation of photovoltaic components.The usual reason of hot spot failures is photovoltaic cell module being blocked partially.For such hot spot failures caused by non-self problems,a set of photovoltaic hot spot failure detection systems capable of real-time monitoring is of great significance for extending the life of photovoltaic cell components and reducing power generation costs.However,the traditional hot spot detection method has low efficiency and low accuracy,which makes it difficult to meet the actual needs of photovoltaic power plants.Based on infrared image recognition technology,a high-efficiency and high-accuracy photovoltaic hot spot detection system has been studied and designed in this paper.The specific work includes the following four points:(1)Discussions about the causes of photovoltaic hot spot failure and its imaging features.Analyzations of the causes of photovoltaic hot spot effects and their possible harm.(2)In view of the blind-spot and low efficiency of manual inspection methods,based on the DJI M100 UAV platform,a small drone autonomous aerial photography application system is designed and developed for efficient aerial photography photovoltaic power stations,which greatly saved the cost of time and labor.(3)In terms of the problem in image positioning caused by the great difference of grayscale in original infrared image after the acquisition,a high-precision segmentation algorithm based on the local grayscale feature of the photovoltaic array area is proposed.Based on the local standard deviation feature,filtering size and position information,supplemented by morphological dilation,the high-precision segmentation algorithm in this paper has achieved accurate segmentations of the photovoltaic array area.(4)Aimed at solution of the problem of low efficiency and accuracy of traditional hot spot detection methods,this paper is proposed to use support vector machines to efficiently detect photovoltaic hot spot failures.The image blocking technology is used to divide the infrared image of the photovoltaic array,and based on the infrared features of photovoltaic hot spots,the temperature features of the sub-image blocks are extracted.SVM is also used to realize the detection of hot spot.The experimental results and the actual measurement of the system show that the infrared spot recognition-based photovoltaic module hot spot detection method adopted in this paper not only realizes the effective identification of hot spot failures,but also has higher detection efficiency and accuracy as comparing with traditional detection methods.This method can meet the actual needs in the work of photovoltaic power plants,and it is positive to reduce the harm of photovoltaic hot spot effect furthermore.