Quantitative Analysis of Aeroengine Turbine Disk Surface Crack under Natural Magnetic Field

Aiming at surface cracks of aeroengine turbine disk, a weak magnetic detection technology without excitation is proposed. Signal processing is performed using blind source separation (BSS). Support vector machine (SVM) is used to realize quantitative evaluation. Four nickel-base superalloy test blocks with defects are designed. The magnetic induction intensity of the test block is collected by a weak magnetic detection instrument. Then, the signal processing based on BSS is carried out. Finally, SVM is used to quantitatively analyze the weak magnetic signal before and after BSS. Results show that SVM estimation accuracy of defect length, width, and depth is 90%, 94%, and 71% after BSS, respectively. These findings are 14%, 8%, and 23% higher than the estimation accuracy prior to BSS.