Detection of Blades Damages in Aero Engine

The health condition of engine which is the key part of modern aircraft determines whether the aircraft can safely carry out the flight mission. As for the detection of the mainstream technology, borescope inspection technology can effectively diagnose the fault and provide a strong guarantee for the next maintenance work. Recently the hardware of borescope inspection technology develops rapidly, but the grade of automation and intelligence is relatively inferior. It is necessary to develop a model detecting flaw parts for borescope equipment because of the lack of the corresponding software system. However, complex conditions inside the engine and various injures types will affect the image quality and increase the difficulty of recognition. In this paper, the YOLOv3 algorithm is applied to the detection task of aero-engine blade damage. Compared with other algorithms, YOLOv3 could have a good tradeoff between detection accuracy and detection speed. In this paper, a network model which could detect the aeroengine fans faults images automatically is trained by train data set. Then, test set is fed into it as follow to verify the validity of detection results, visual analysis is also performed on these results. Finally, our research method demonstrates us a satisfied recognition results of blade damages.