Classification of internal defects of gas turbine blades based on the discrimination of linear attenuation coefficients

Abstract

In the process of manufacturing and servicing gas turbine blades, various types of defect are formed and grow rapidly due to the extremely harsh working environment, which poses a huge threat to the safe operation of the gas turbines. Given that different types of defect cause varying degrees of damage to the turbine blades, it is vital to distinguish and deal with defects differently. Considering the shape of the blade (free-form surface) and the location of the defect (inside the blade), digital radiographic imaging can be used for the non-destructive testing of turbine blades. Although some types of defect (for example porosity and cracks) can be distinguished from others (for example voids and inclusions) based on differences in morphological and textural characteristics, others (for example voids and inclusions) may be misclassified due to similarities in morphological and textural characteristics. These defects with similar morphological characteristics are composed of different materials, which can be utilised as a basis for classification. This paper presents a classification method for defects with similar morphological characteristics based on the discrimination of linear attenuation coefficients. Several typical defects, including voids and inclusions, are set into a cuboidal block and into nylon blades in this work. Their corresponding linear attenuation coefficients are obtained. A binary classification of the linear attenuation coefficient enables the categorisation of voids and inclusions. Experimental results demonstrate that the proposed method has high efficiency and the judgement for voids and inclusions is accurate.