A new modified incomprehensible but intelligible-in-time logics algorithm for modeling solid oxide fuel cell.

Area of Science:

• Electrical Engineering
• Computational Intelligence
• Chemical Engineering

Background:

• Accurate equivalent circuit modeling is crucial for solid oxide fuel cell (SOFC) performance evaluation.
• Optimization algorithms are essential for parameter estimation in complex systems like SOFCs.
• Existing algorithms may struggle with local optima and balancing exploration/exploitation phases.

Purpose of the Study:

• To introduce a novel optimization algorithm, REILA (Randomized Exploration and Intelligent Learning Algorithm), for enhanced parameter estimation.
• To apply REILA for determining the optimal parameters of SOFC equivalent circuits.
• To validate REILA's performance against established algorithms on benchmark functions and engineering problems.

Main Methods:

• Development of the REILA algorithm incorporating a random contraction strategy (RCS) for exploration and a new exploitation phase.
• Application of REILA to IEEE CEC'2022 test functions and three constrained engineering problems.
• Evaluation of REILA's effectiveness in mitigating sum mean squared error (SMSE) for SOFC steady-state and dynamic models.

Main Results:

• REILA demonstrated superior performance compared to nine well-known algorithms on benchmark functions.
• The algorithm achieved a minimal fitness function of 1.5528e-06 for the SOFC steady-state model at 1173 K.
• The lowest error for the dynamic SOFC model was found to be 1.1267 using REILA.

Conclusions:

• REILA effectively avoids local solutions and achieves a good balance between exploration and exploitation.
• The proposed method is suitable, resilient, and proficient for evaluating SOFC equivalent circuit parameters.
• REILA shows significant potential for solving real-world engineering optimization problems.