Fault diagnosis of SOFC system based on single cell voltage analysis

The commercialization of Solid Oxide Fuel Cell (SOFC) systems is restricted by the low reliability and durability severely. Generally, a SOFC system includes stack, combustion chamber, fuel heat exchanger, air heat exchanger, steam generator, reformer, and blower, etc. Faults of any part of the SOFC system will affect the performance of the entire system at any time, causing a decrease in the durability and reliability. Therefore, more and more attention has focused on the fault diagnosis technology on the SOFC system lifetime research. However, the practicability of current fault diagnosis algorithm is not enough on account of the redundant fault signals. This paper proposed an improved algorithm for the fault diagnosis using the single cell voltage as the only fault characteristic signal. The voltage signal in time domain with four representative system faults (stack degradation failure, reformer degradation failure, fuel leakage failure, and air leakage failure) are generated by simulation. The fault voltage signal in frequency domain is obtained by Fourier transform of voltage signal in time domain. Then the characteristics of the fault voltage signal in time domain and in frequency domain are extracted, and the voltage performance of fault signal in time domain are analyzed. The single fault and simultaneous fault were diagnosed for four representative system faults and their combination using a suitable classifier. Both types of the fault diagnoses have achieved good recognition effects, which fully verified the feasibility of this improved algorithm. (c) 2021 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Automated summary

The commercialization of Solid Oxide Fuel Cell (SOFC) systems is restricted by low reliability and durability issues, leading to a focus on fault diagnosis technology for extending system lifetime. This paper proposes an improved algorithm using single cell voltage as the only fault characteristic signal, achieving good recognition effects for diagnosing four representative system faults and their combination.