基于多层对比学习的航空发动机内部损伤智能检测方法OA

When detecting internal damage in aircraft engines,single level features are difficult to accurately reflect the internal damage status of the engine,resulting in poor damage detection accuracy of the method.In this regard,an intelligent detection method for aeroengine internal damage based on multi-layer comparative learning is proposed.The feature extraction network in multi-layer contrastive learning is applied to extract multi-level features from the internal images of aircraft engines.Then,a projection head is designed to map the features to a low dimensional contrastive learning space,construct feature vectors,set a loss function,and combine with a segmentation head to segment the damaged areas of the internal images of the engine.The texture characteristics and geometric characteristics are extracted from the segmented damage area,the damage degree of the engine internal damage area is calculated,the damage level is divided,and the engine internal damage detection is realized.Experimental results show that the fit between the test results and the actual results in practical applications is 95.36%,and it has high detection accuracy.