Abstract: Corrosion defects in the buffer layer of high-voltage cable metal sheaths are characterized by high latency and strong concealment, significantly compromising cable reliability and safety. To address these defects, an integrated detection method based on multiple feature parameters was proposed. The method integrated infrared detection and X-ray imaging to extract multiple feature parameters. Differential evolution algorithm was employed to optimize the parameters of the K-Means clustering algorithm. Using the multi-feature parameters as input, defect identification was accomplished through K-Means clustering. The results show that compared with three traditional detection methods, the proposed method achieved the lowest log loss, indicating it had the highest accuracy in defect detection.