Study on Model Evolution Method Based on the Hybrid Modeling Technology With Support Vector Machine for a SOFC-GT System

Abstract

Abstract The mechanism models of solid oxide fuel cell–gas turbine (SOFC-GT) systems are very useful to analyze the detail thermodynamic performance, including the internal complex mass, heat and electrochemical processes. However, several characteristic parameters in the mechanism model are difficult to be estimated accurately due to the unknown offset. As a result, it is difficult for the mechanism model to maintain high accuracy during the full operating cycle. In this paper, a model evolution method based on hybrid modeling technology is proposed to simulate the thermodynamic performance more accurately during the full operation cycle. A hybrid model framework of SOFC-GT system is designed to evolve the mechanism model. The electrochemical characteristic of SOFC is identified and evolved by a data-driven model based on least squares-support vector machine algorithm (LS-SVM) rather than a mechanism electrochemical model. Firstly, the prediction performance of the electrochemical LS-SVM model is compared with the test data. The maximum error of prediction is only about 1.776 A/m2, and the prediction accuracy reaches 99.998%. Then the hybrid model, coupled with the LS-SVM electrochemical model from the mechanism model, is developed to simulate the thermodynamic performance of SOFC-GT system. The off-design performance of the SOFC-GT system is analyzed by the hybrid model and mechanism model. In addition, the comparison results show that the hybrid model can accurately predict the SOFC-GT system performance. The maximum error is less than 2.2% at off-design condition. In consideration of its significant advantages combining data-driven model and mechanism model, hybrid model is a powerful candidate for accurate performance simulation during full operation cycle.