Quantitative characterization of microstructures and elastic properties in plasma sprayed coatings based on "Ultrasonic - Elasticity - Microstructure" interaction mechanism

In plasma sprayed coatings, pores are the primary microstructure, and their features such as porosity, shape, and orientation significantly influence the thermal insulation and elastic properties of the coatings. Therefore, the ultrasonic characterization of pore feature parameters plays a crucial role in guiding the processing and predicting the lifespan of coatings. However, the coupling of multiple pore feature parameters leads to ill-posed problem and nonlinearity between these parameters and the ultrasonic feature signals, making it challenging to identify multiple pore feature parameters accurately. In this study, based on the "ultrasonic - elasticity structure" interaction mechanism, a novel strategy for the ultrasonic quantitative identification of pore feature parameters and elastic constants in plasma sprayed coatings is proposed.Firstly, a random pore model (RPM) was developed to reveal the complex morphology of pores in plasma sprayed coatings. The porosity, average aspect ratio, orientation factor, and contour roughness factor were statistically analyzed for 15 sets of constructed RPMs, followed by ultrasonic finite element simulations. Then, a sensitivity matrix (SM) inversion method is developed for the "ultrasonic - elasticity" relationship by accurately measuring the elastic constants of the coatings through ultrasonic wave velocity inversion at multiple angles, yielding a relative error of only 0.57%. Lastly, a Multi-output Support Vector Regression machine learning model optimized by genetic algorithm (GA-MSVR) was proposed to establish the "elasticity - microstructure" relationship and identify multiple pore feature parameters based on elastic constants. The maximum relative errors for the four pore feature parameters are found to be 12.28%, 6.35%, 18.87%, and 7.62%, respectively, demonstrating the feasibility of the proposed strategy. Furthermore, the sensitivity matrix was employed to explain the cause for the larger relative error of the orientation factor.In summary, a challenging task of quantitatively identifying pore feature parameters in plasma sprayed coatings is addressed by proposing a novel solving strategy based on the "ultrasonic - elasticity - microstructure" interaction mechanism. The developed RPM, SM inversion method, and GA-MSVR model contribute to the accurate and comprehensive identification of pore feature parameters. The results obtained validate the feasibility of the proposed strategy.