Optimization of energy density in supercapacitors by utilizing a hybrid artificial neural networks-genetic algorithm based optimization algorithm

Energy storage systems are required to meet the increased energy demand and reduce the need for fossil fuels. Energy storage devices are of interest to renewable energy systems and electric vehicles to provide a permanent energy supply. Supercapacitors, in particular, are eligible systems for energy storage owing to their unique properties such as very long cycle life, high reversibility, and high power density. Nonetheless, they have limited energy density. The ultimate goal is to increase the energy density of supercapacitors while maintaining high power density. In this thesis, hybrid artificial neural network-genetic algorithm (ANN-GA) model is utilized to increase the capacitance of supercapacitors. Several data preprocessing, feature selection, and machine learning algorithms are performed to predict the capacitance of supercapacitor by using experimental data. It is observed that ANN is a powerful method to capture nonlinear relationships concerning the physical and operational features of supercapacitors. GA is a promising method that examines search space for the optimal solution. The trained neural network model is used as the fitness function for genetic algorithm to achieve maximum capacitance within the feasible range. Selection, crossover, and mutation procedures are implemented in the reproduction step of GA to offer elaborate search space. In a nut shell, this study takes a step towards the rational design of supercapacitors by implementing a hybrid ANN-GA as an optimization tool to improve the capacitance. The results indicate that obtained optimal design parameters agree with the literature while improving the capacitance of supercapacitors significantly.