| AlSi-polyester seal coating has many advantages,such as low friction coefficient,strong anti-erosion property,stable performance,easy adjustment and maintenance,and has been widely used in aircraft engines.In an ideal seal coating,the distribution of non-metal lubricant and pores are uniform in the metal matrix.In order to balance the abradability and the erosion resistance,the composition design and microstructure regulation should be carried out.Therefore,accurate nondestructive characterization of porosity and microstructure uniformity is an effective way to fundamentally control coating quality and improve its service performance.In nondestructive characterization of microstructure,ultrasonic testing technique has the incomparable advantage over other methods,but the ultrasonic nondestructive characterization of seal coating work is still in its infancy.In particular,the problems and challenges are as follows:the non-metal and pores in seal coatings have the feature of irregular shape,large range of size,and random distribution on the microscopic scale.The honeycomb structure and the randomness lead to strong fluctuation in local elastic properties of coating.When ultrasonic wave propagates in the coating,Rayleigh scattering,random scattering,multiple scattering and other scattering mechanisms will occur.In the frequency domain,the ultrasonic attenuation is nonlinear and shows an obvious ’multi-scale scattering effect’.The empirical fitting methods and regularization methods are difficult to accurately characterize the porosity and microstructure of the coating.In order to solve the above problems,a random multi-phase medium modeling method was established to describe the microstructure of coating.Combined with the numerical simulation and ultrasonic scattering theory,the attenuation mechanisms of pore,polyester and multi-factor coupling were preliminarily elucidated.Furthermore,a Support Vector Regression model optimized by Particle Swarm Optimization algorithm(PSO-SVR)based on multi-scale ultrasonic attenuation coefficient was further established.Finally,the quantitative nondestructive characterization of the porosity and microstructure uniformity of the AlSi-polyester seal coating were achieved.The main research contents and conclusions are listed as follows:(1)According to the spatial distribution of complex coating elastic properties,existing ultrasonic models are difficult to describe the complex cellular structure and the inherent randomness flexibly and accurately,and can not actually reflect complex ultrasonic propagation behavior insidecoating.Based on the random medium theory combined with statistical methods,in this paper,a Random Multi-phase Medium Model(RMMM)for heterogeneous coatings containing solid-solid-gas three phases is developed,which is able to simulate the elastic spatial distribution of coatings by taking into account material microstructure.It is also applied to the analysis of ultrasonic propagation characteristics and acoustic simulation.Results indicate that the random polyphase model can effectively reconstruct the microstructure of the coating and fully reflect the complex morphology of polyester,pore and the inherent randomness of the coating.When porosity increases from 0.3%to 4.0%,the longitudinal wave velocity decreases from 2251 m/s to 2150 m/s whose relative variation ratio is about 4.5%,while the attenuation coefficient increases from 4.45 dB/mm to 6.56 dB/mm whose relative variation ratio is up to 47.4%.The simulated results show a good consistency with those of the experiments.(2)In view of the traditional heterogeneous medium attenuation mechanism failed to system in-depth analysis in the study of phase type,content,size,distribution of their respective functions,especially how the coupling influence between acoustic attenuation characteristics of the problem,this paper embarks from the coating inherent heterogeneous heterogeneity characteristics,based on the analysis of "split-integration" strategy.Numerical simulation of random multiphase medium model is used to clarify the influence of component phase type,content and other single factor conditions and multi-factor coupling on the sound attenuation characteristics of the coating.The subregional approximation theory of ultrasonic scattering and the generalized self-consistent theory are combined to establish the analysis model of the sound attenuation mechanism of the multiphase non-uniform sealing coating.First of all,based on the typical AlSi-polyester seal coating composition content and microstructure characteristics of the statistical results,the full-component random multiphase medium is divided into ultrasonic models containing only pore and only polyester.The effects of content,size,distribution uniformity and multi-scattering of pore and polyester on the coating’s attenuation characteristics are revealed by numerical simulation with acoustic scattering theory.Then,the attenuation correction term of the interaction between pore and polyester is introduced through full-component numerical simulation.Finally,the above analysis is integrated to clarify the complex sound attenuation mechanism of the AlSi-polyester seal coating under multi-factor coupling constraints.(3)In view of the nonlinear and ill-posed problems in the inversion of porosity and micro structure uniformity parameters,this paper proposes a PSO-SVR prediction model based on multi-scale ultrasonic attenuation coefficient by combining wavelet transform multi-scale analysis and machine learning algorithm from the perspective of ultrasonic feature parameter extraction and coating parameter inversion.The porosity and microstructure uniformity of the AlSi-polyester seal coating were predicted using the new model.The results showed that,the R2 and MSE between the predicted porosity of the model and the in situ calibration values of the micro CT were 0.947 and 0.067 respectively,and the R2 and MSE between the predicted uniformity length of the model and the in situ calibration values of the micro CT were 0.834 and 0.824 respectively.Compared with the time domain and frequency domain analysis methods,the prediction performance of PSO-SVR model with multi-scale attenuation coefficient spectrum as the input feature space is significantly better than that of frequency domain attenuation spectrum and time domain attenuation coefficient.The comparison results show that the PSO-SVR model has higher accuracy,better generalization ability,and stronger robustness compared with classical ANN models,in the case with limited experimental data.As an extension of this work,the PSO-SVR model is expected to be applied to the ultrasonic quantitative nondestructive characterization of other multi-phase heterogeneous materials. |
PDF Full Text Request