A Damage Assessment System for Aero-engine Borscopic Inspection Based on Support Vector Machines

Defects are often arise on the inner surface of an aeroengine, but most of the aeroengine borescopes can only detect the damages and cannot determine the degree of damages. We propose a novel borescope assessment expert system (ES) to evaluate the degree of typical flaws of an engine and to provide the corresponding maintenance advices. The system put typical damage images and relevant maintenance rules into knowledge bases as the standard cases. A binary-tree-based support vectors machine (SVM) was used as the reasoning machine to obtain case knowledge and implement the logic reasoning, which enhanced the learning ability, inference speed and precision of the expert system. The application to CFM56 aero-engine shows that the system with both the advantages of SVM and ES has higher assessing accuracy than traditional ES method.