Research And Application On Solid Oxide Fuel Cell Microstructure Image Segmentation

The characterization of the microstructure and its related electrical properties of porous electrode are the necessary to optimize the optimum fabrication conditions for the best performance of SOFC.Currently the microstructure image analysis method is still in immature stage.Hence,we considered that three-phase of electrode under optical microscope presents the specific distribution regulation and exists interference factors.Accurate three-phase recognization provides algorithm support of microstructure parameters output.The quantum-inspired adaptive fuzziness factor is defined to adaptively estimate the energy function in the fuzzy system based on MRF.Before defuzzification,a quantum-inspired probability distribution based on distance and gray correction is proposed,which can adjust the inaccurate probability estimation of uncertain points caused by noises and edge points.It is an ingenious integration of quantum properties with probability estimation during clustering and solve the problems existed in other clusterting algorithms,such as noise suppression,image details preservation.Due to the insufficient noise supression ability in traditional GMM,a coarseness-entropy adaptive factor is defined to automatically incorporate the spatial contextual information into iteration formula of EM algorithm,which can control the trade-off between robustness to noise and effectiveness of preserving the details.Thus the internal clustering probability constraint theory based on the combination of coarseness-entropy and spatial contextual information is created.The proposed method can solve the problems of inhomogeneity and low accuracy of segmentation results.Finally,applying to simulated images and actual electrode images,the two proposed methods obtained expected effects.SOFC image analysis technical is improved and optimized by using above theories.We can quantify the microstructure parameters as image segmenation results input of images.It provides data support and theriocal provement to optimize the process of fuel cells prepatation.