To address the issues of defect leakage and mis-segmentation in the surface defect segmentation of solid oxide fuel cells, we propose a cascaded mixed pyramid pooling segmentation method. In the encoding part, attention-based depth-wise separable convolution is implemented to replace the traditional convolution. This reduces the computational complexity and achieves hierarchical feature extraction. Secondly, cascaded sampling is applied to the last three layers of the encoder to promote multi-level feature fusion, and mixed pyramid pooling consisting of atrous convolution and strip pooling is adopted to achieve feature extraction for long-range band defects. Lastly, an adaptive loss function is constructed to supervise the training process, balancing the relative importance among the losses and improving the learning of defect features by the network. The experiments demonstrate that the proposed method improves segmentation accuracy, as evidenced by better results in mean absolute error, F1 score, intersection over union, and Pratt's figure of merit.
One of the main goals of material design is to sift the proper materials with the properties we want. However, the traditional method, synthesizing and testing each material in laboratory, wastes time and energy, and the actual material we want is usually one in a million which makes it more difficult. Here, we develop a generative framework to give a guidance on material design with specific properties. Our framework is mainly drove by several variants of Generative Adversarial Networks (GANs) for material data generation. Our framework is trained with 86 perovskite-type material samples including their components information, and then we compared with various networks structures and algorithm, the result shows an acceptable accuracy of materials data generation which proved a possible method of inverse design of perovskite-type electrode of SOFC.
Three-phase segmentation of solid oxide fuel cell (SOFC) anode image is essential for its microstructure quantitative analysis. A quantum-inspired mixture clustering model is developed for Ni/YSZ anode optical microscopy (OM) images. The proposed mixture clustering model focuses on combining Markov random field (MRF)-based fuzzy logic model with Gaussian mixture model (GMM). In the premise part of fuzzy if-then rules, the clique potential MRF function is defined by multiplication of both prior distributions and fuzzy membership functions with a quantum-inspired adaptive fuzzy degree, which takes prior statistics properties and spatial contextual information of SOFC anode OM image into consideration. In addition, GMM is introduced into the consequent part to design a negative log-prior function as the pixel distance metric. The proposed method has been compared to other state-of-the-art segmentation algorithms on both simulated images and real SOFC anode OM images. The experimental results demonstrate that the proposed method is able to achieve a higher segmentation accuracy with a faster convergence speed, which lays firm foundation for microstructural parameters extraction from SOFC electrodes image datasets.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.