Power Output Prediction Method for Fuel Cell Based on Improved Genetic Algorithm Optimized Back Propagation Neural Network Model

Solid oxide fuel cells (SOFC) possess efficient and clean energy conversion characteristics, enabling the direct conversion of various fuels into electricity, making them a vital technology in the future clean energy sector. This paper proposes a prediction method based on an improved genetic algorithm optimized neural network (GGANN) model to address the challenges in predicting power output of SOFC systems. This paper has enhanced the selection, crossover, and mutation operations of genetic algorithms to favor individuals with higher fitness, precisely explore solution spaces, and augment population diversity, thus mitigating premature convergence. Through simulation of SOFC system operation using model to obtain key parameter data, the iGANN model is trained. Comparative analysis with back-propagation neural network (BP), genetic algorithm optimized neural network (GANN), and long short-term memory (LSTM) demonstrates the superior predictive accuracy of the proposed model, showing significant improvement over traditional models. This paper provides new perspectives and methods for the design and optimization of SOFC systems, playing a crucial role in advancing clean energy technologies.