Measurement Science and Technology · 2021 · 10 citations · 27 references
EngineeringMaterials EngineeringMechanical EngineeringComputer VisionHeat TransferMachine VisionDigital Image ProcessingImage AnalysisNon-destructive Crack EvaluationNon-destructive InspectionNon-destructive TestingNondestructive EvaluationThermal Imaging In InspectionThermal Imaging InspectionThermal Infrared ImagingThreshold-based Segmentation
Abstract Thermal barrier coatings (TBCs) are widely used in gas turbines. Crack is one of the most serious failure modes in TBCs. In this work, an investigation of thermal imaging in non-destructive tests to detect cracks in the inner volume of TBCs has been carried out. A novel algorithm based on gray gradient space histogram entropy (GGSHE), multi-threshold segmentation and stacked restricted Boltzmann machine (GEMSS) is proposed. GGSHE extracts the color and spatial features from the acquired infrared image, and then multi-threshold segmentation is used to segment the crack tentatively. A stacked restricted Boltzmann machine is adopted to diagnose the crack length as well as the position and density. Experiments are conducted to test the performance of GEMSS and other algorithms. In the theoretical experiments, it is found that the average accuracy of the proposed algorithm is 95.5% under crack position detection and 86.7% under crack density detection. Theoretical experimental results also reveal that GEMSS can achieve the highest F -score (0.85) under crack length evaluation among the algorithms. Experiments on the effect of image noise on the detection of the algorithm are also carried out and discussed. Coatings with different prepared crack lengths are fabricated and tested using our algorithm. Based on the results of this study, the proposed method shows good potential to evaluate internal cracks of TBCs in actual process.
27