Implementation of a low-code monitoring system for solid oxide fuel cells

Abstract

Solid oxide fuel cells (SOFCs) have the advantages of high conversion efficiency, low working noise, and environmental protection. Thus, they are widely used in large-scale power generation, combined heat power, peak load-regulating energy storage, and other fields. In the actual working process of SOFCs, the battery performance will decline sharply due to extreme conditions such as overtemperature and peroxide, and the service life will be terminated. Remote monitoring of fuel cells can detect problems in time and restore the battery's performance by restoring the battery, thereby extending its service life. The data generated by fuel cells during operation have a strong temporal correlation. To discover the future state of fuel cells in time and solve the problem that a large number of codes are needed for monitoring, this paper designs a remote monitoring system of fuel cells with low-code based on Node-RED and Ali Cloud. The output current of fuel cells is predicted by using the improved back propagation neural network based on Levenberg-Marguardt (LM-BPNN) algorithm. The test result of LM-BPNN shows that the test samples have a root mean square error of 0.172, a mean absolute percentage error of 3.85%, and a coefficient of determination of 0.998.