Data Fusion for Combining Techniques to In Situ Detect Defects of Turbine Blade

摘要

In non-destructive testing (NDT) it is important to have a high probability of detection (POD) and reliable characterization of each defect. This can be gained by using several techniques, which leads to an increase in cost and time for testing. Another option is to use several techniques combined into a single probe, in which case data fusion for the techniques is possible. A dual probe containing an eddy current probe and a borescope is presented. These two techniques are complementary, and no detrimental interference is observed. This dual probe can in situ detect defects of complicated object such as aeroengine, which can reduce the need for costly teardown. Turbine blade is one of crucial parts in aeroengine. Defects of turbine blade would endanger flight security. Cracks and corrosion are primary defects of turbine blade. Cracks and corrosion are detected in situ using eddy current and borescope simultaneity. Based on the Dempster-Shafer (D-S) theory of evidence, a decision level data fusion was used to combine probability mass values from borescope and eddy current. The final classification results were obtained by making decisions based on the maximum belief of the fused results. The experimental results show that the high reliable characterization of defect is gained using the data fusion.