Processing And Fusion Analysis Of Multi Field Measurement Data Of Typical Swirling Flame

Swirl flame in confined space has a strong demand for industrial application,nowadays its theoretical description and fine experimental research is still insufficient.The vortex breaking process induced by cyclonic flow plays an important role in flame stability,but the physical nature of flame instability,thermoacoustic oscillation and tempering,especially the coupling mechanism of turbulence and combustion,has not been thoroughly studied.In this paper,the experimental study of OH density and velocity field measurement of swirling flame is carried out,the data images are de-noised and super-resolution reconstructed,and the preliminary fusion analysis of OH density and velocity field are carried out.The results show that the combination of image denoising and neural network method is feasible for super-resolution reconstruction of OH density segmentation image of swirling flame,which is helpful for fusion analysis of OH density segmentation and velocity field.Firstly,the OH-PLIF experiment,the contra-side OH-PLIF synchronous imaging experiment and the OH-PLIF and PIV synchronous measurement experiment were described in detail in this paper.The data for image denoising,super-resolution reconstruction and fusion analysis of the OH density and velocity field were obtained.Secondly,the image preprocessing methods to enhance the boundary and structure information of OH density field images are analyzed.The peak signal-to-noise ratio（PSNR）and similarity structure（SSIM）are introduced to evaluate image quality,and the effects of seven image denoising algorithms,including Gaussian low-pass filtering,mean filtering and median filtering,on improving the image quality of swirling flame were compared.The results show Gaussian low-pass filtering,median filtering and mean filtering algorithms have the best denoising effect on the OH density of swirling flame images with low SNR,and the gaussian low-pass filtering algorithm is found to be more effective in enhancing the boundary features of swirling flame images.Thirdly,the study of image super-resolution reconstruction based on SRCNN method was carried out using the OH-PLIF synchronous imaging data on the opposite side.Compared with the original low-resolution images,the reconstructed results improved by 5d B in PSNR index and greatly improved in boundary clarity.In the last part of this work,the data fusion of OH-PLIF and PIV synchronization velocity field and the OH density field of swirl flame has carried out by using the images of different cross section under different cases and with the change of time.We believe the work would help to further understand the mechanism of interaction of flame velocity and composition field.