Visualization of Damage Progress in Solid Oxide Fuel Cells

Abstract

The fuel cell is regarded as a highly efficient, low-pollution power generation system. In particular, Solid Oxide Fuel Cell (SOFC) has a high generation efficiency. However, a crucial issue in putting SOFC to practical use is the establishment of a technique for evaluating the deterioration. We previously developed a technique by which to measure the mechanical damage of SOFC using the Acoustic Emission (AE) method. In the present paper, we applied the kernel Self-Organizing Map (SOM), which is an extended neural network model, to produce a cluster map reflecting the similarity of AE events. The obtained map visualized the change in occurrence patterns of similar AE events, revealing four phases of damage progress. The methodology of the present study provides a common foundation for a comprehensive damage evaluation system and a damage monitoring system.

Concept Drift Detection with Clustering via Statistical Change Detection Methods

“…To handle high-frequency sensor data streams and to cluster AE events together with annotations, we propose the architecture shown in Fig. 1, which combines a data stream management system (DSMS) [12], [13] and a data mining method [2] in cooperation with a user. The figure shows the two types of processing.…”

“…Here, Pr(S 1 ) gradually decreases and Pr(S 2 ) increases over time. In SOFC monitoring, damage types change over time depending on the inner state [2], for example, damage gradually shifts from cracks in the electrolyte to cracks in the glass seal.…”

“…2c, incremental drift is a gradual change of Pr(X|S) that becomes a different distribution of variable X due to changes in the nature of S. This type of drift is typically the most difficult to detect. In SOFC monitoring, the characteristic power spectrum of AE events, even from the same material, changes depending on the temperature and oxidation stage [2].…”

“…Such damage is observed via ultrasonic waves of acoustic emission (AE) produced by cracks and/or delamination of the materials. Fukui et al validated that major types of damage can be clustered on AE events via Kernelized self-organizing maps (SOM) [2]. Here it is impractical to collect all possible damage types in advance of monitoring; therefore, it is important to renew the cluster model while adapting to changes in characteristics and types of damage, i.e., concept drift [1], [3].…”

Statistical sleep pattern modelling for sleep quality assessment based on sound events

“…However, one of the problems of their method is cluster detection, even they could find the best match unit on the cluster map for every input data, it is difficult to find out the major clusters on the cluster map without manually annotation of the input data, which will cost lots of time. In this paper, we applied HC on the cluster map from Kullback-Leibler kernel SOM (KL-KSOM) [13], HC is a method that seeks to build a hierarchy of clusters, builds nested clusters by merging or splitting them successively. We calculated the distances between cells on cluster map, and detected the hierarchical structure of cells by HC.…”