High Quality Imaging Research On Turbine Blades For Knowledge-Based Cone Beam CT

Superalloy hollow turbine blades are hailed as the “crown jewels” of aero-engines and are key components that affect engine performance and safety,whose casting accuracy and inspection are particularly important.If there are defects such as micro-cracking,uneven wall thickness,or out-of-range size in the turbine blades,the service performance of the blades will be greatly reduced,and even serious accidents such as engine explosions will occur.Therefore,the size detection and non-destructive testing(NDT)for turbine blade is the urgent and key problems need to be solved.The research on fast and effective NDT and high-quality imaging is critical to solving turbine blade inspection and engine development.Due to the scattering and beam hardening who has affected the imaging quality and made it poor in aeroengine turbine blade inspections,in this paper,the problems of noise,scattering,beam hardening artifact correction and ill-posed projection reconstruction in CBCT system imaging process are studied.Specifically,CBCT scanning imaging analysis,scattering field estimation,beam hardening correction based on curve modeling,blade feature detection,and ill-posed projection reconstruction algorithm are performed.The main research content includes:1.For the imaging quality problem of turbine blades,the characterization of the entire CBCT projection imaging process is given.At the same time,the influencing factors in the imaging process of the turbine blades are described.The imaging knowledge of the CT system and the characteristics of the turbine blades are analyzed.Based on the mechanism of imaging noise and artifact generation,the effects of exposure parameters,quantum noise,detector response and output on the imaging of turbine blades were discussed.Combining with the characteristics of turbine blades,the characteristics of blade projection and reconstruction artifacts are analyzed,and the processing methods for suppressing imaging noise and artifacts are given.Finally,according to the impact of the entire projection imaging chain and each link on the imaging quality,a related strategy to achieve high-quality CBCT imaging is proposed.2.Aiming at the problem of scattering correction of turbine blades in CBCT imaging,a scattering estimation method combining grating scattering with structural tensor region feature modeling is proposed.The method uses the gradient information of single-scatter scattering detection and blade projection as a priori knowledge,divides and models the feature region of the projected image,and combines the local angular position grating scattering detection information to obtain the knowledge of the projection-scattering model parameters of different regions.The estimation of the scattered field based on the angle sequence projected image is completed.The experimental results show that the scattering field obtained by the feature of the structure region can effectively solve the problem of the scattering field estimation of the angle sequence.Finally,by comparing different scattering acquisition methods,it is concluded that the scattering estimates in the structural region are closer to the grating scattering field distribution.3.To solve the effects of beam hardening and scattering coupling artifacts on imaging quality in the imaging process of turbine blades,firstly,the path hardening knowledge of radiation is completed based on the behavior of beam hardening,and a method based on binarization is proposed.Then the DR projection is corrected for scattering in the projection domain and the slice domain.By analyzing and comparing the results,the beam hardening correction was completed.The experimental results show that this method effectively suppresses the influence of the scattered field and beam hardening artifacts on the reconstructed image.After the scatter correction,the scattering artifacts are reduced,the signal-to-noise ratio is raised above 10%,the average gradient is raised above 80%;and after the beam hardening correction,the cup artifacts are also well suppressed,which has improved the imaging quality,effectively.4.For the problem of ill-posed projection reconstruction problem,the ill-posed features are first detected to obtain the blade feature knowledge,then the pre-processing methods for ill-posed projection are given.Aiming at low-signal projection,a method of local projection multiple averaging is proposed to improve the signal-to-noise ratio.In view of the impenetrable projection of rays,based on the improved discrete algebraic reconstruction algorithm,a hybrid fusion anti-noise reconstruction algorithm is proposed.This method combines the algebraic iterative algorithm with the FDK algorithm for updating the slice.The experimental results show that the normalized mean square distance(NMSD),normal average absolute distance(NAAD),and signal-to-noise ratio(SNR)are all improved by more than 10% for the reconstructed slice images.