Abstract: Under the influence of factors such as interference from the marine environment, reflection and radiation characteristics on the surface of submarine cables, there is redundancy in temperature distribution information, leading to problems such as high LOSS loss function values and incomplete identification of fault areas in the process of identifying faults in submarine high-voltage optoelectronic composite cables. Therefore, a long-distance and ultra large field of view infrared identification method for submarine high-voltage optoelectronic composite cable faults based on thermal path analysis function is proposed. By using the layered functions of the surrounding environment and the main environment of the submarine cable, a thermal path analysis function for the submarine high-voltage optoelectronic composite cable is constructed to generate temperature field information of the submarine high-voltage optoelectronic composite cable. Through the infrared radiance conversion method, the infrared image of the submarine cable is converted to obtain the linear relationship between the infrared image fields of view. The YOLOv3 function is linearized and improved by combining the K-means clustering method. The YOLOv3 function is applied to predict the fault coverage area, measure and locate the fault point, and complete long-distance and ultra large field of view fault recognition. The test results of the numerical examples show that the proposed method for identifying faults in infrared images of submarine high-voltage optoelectronic composite cables exhibits high information entropy, edge intensity, average gradient, and spatial frequency values. The loss function value of LOSS continuously decreases, and the fault area of the submarine cable can be fully identified. It can cope with long-distance and ultra large field of view environments, meeting the needs of fault identification for underwater high-voltage optoelectronic composite cables.