Optimal identification of solid oxide fuel cell parameters using a competitive hybrid differential evolution and Jaya algorithm

This paper presents a hybrid metaheuristic optimizer called CHDJ to identify accurate and reliable values for the model parameters of solid oxide fuel cells (SOFC). CHDJ is a hybridizer of differential evolution (DE) and Jaya algorithm. The core of CHDJ lies in a developed competitive hybrid operation. In this operation, three vectors including the DE mutant vector, the Jaya mutant vector, and the crossover vector that constituted by the foregoing two mutant vectors form a tri-competitive mechanism together. This mechanism can contribute to exploring the global space extensively and exploiting the local region meticulously to generate promising candidate solutions. To verify the performance of CHDJ, two SOFC models including a cylindrical cell and a tubular stack operating at different pressures and temperatures are employed. Experimental results demonstrate that these three vectors interact and cooperate organically. They make CHDJ take full advantages of the exploration of DE and the exploitation of Jaya effectively. The optimal values of the mean square error (MSE) between the measured voltage and the calculated voltage obtained by CHDJ on the cylindrical cell at 1073K, 1173K, 1213K, and 1273K are 2.9215E-06, 1.4610E-06, 1.9093E-06, and 2.4682E-06, respectively. On the tubular stack, for the operating conditions of 3atm constant pressure at 873K, 923K, 973K, 1023K, and 1073K, the optimal MSE values are 2.7489E-05, 1.6169E-04, 6.9207E-04, 1.4833E-03, and 1.5876E-03, respectively. For the operating conditions of 1073K constant temperature under 1atm, 2atm, 3atm, 4atm, and 5atm, the results are 1.5293E-03, 1.6289E-03, 1.5876E-03, 1.5868E-03, and 1.5870E-03, respectively. Comparisons with other algorithms show that CHDJ achieves better overall performance. It can skip local optima and find optimal solutions, and thus can serve as a hopeful candidate method for parameter identification of SOFC models. (C) 2020 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Automated summary

CHDJ is a hybrid metaheuristic optimizer that effectively identifies accurate values for the model parameters of solid oxide fuel cells. It utilizes competitive hybrid operations to explore global space and exploit local regions, generating promising candidate solutions. Experimental results demonstrate that CHDJ achieves good performance under different operating conditions.