PHOTOVOLTAIC MODULE HOT SPOT FAULT DETECTION BASED ON HIGH-ORDER SPATIAL INTERACTION NETWORK
Aiming at the problem of traditional infrared hot spot fault detection algorithm is prone to complex background interference and the low fault detection accuracy of small and dense targets,a photovoltaic module hot spot fault detection network based on high-order spatial interaction is proposed.Firstly,the high-order spatial interaction module is designed,and the YOLOv5 backbone network is introduced for global interaction modeling to improve the detection accuracy of the network.Secondly,in order to highlight the key features of fault targets in complex background,a CCA module is constructed based on the CA to reconstruct the neck network.Then,AFFM module in the neck network is designed to enhance the detection accuracy of the detection network of multi-scales.Finally,the self-adaptive feature fusion detection head is designed to improve the model's perception of small targets.The experimental results suggest that compared with the seven classical detection algorithms,the proposed algorithm has the highest detection accuracy,reaching 84.3%.
PV modulesfault detectiondeep learninghot spot effecthigh-order space interactionfeature fusion
NETL
NSTL
万方数据
