Energy, exergy, economy analysis, and multi‐objective optimization of a novel integrated energy system by combining artificial neural network and whale optimization algorithm

This study proposed a novel solid oxide fuel cell (SOFC)-integrated system fueled by biogas production from anaerobic fermentation of organic municipal solid wastes. The proposed system composed of the organic Rankine cycle (ORC), the anaerobic digester (AD), the SOFC, Kalina cycle (KC), and the parabolic trough solar collector (PTSC) which provided heat for the continuous and stable operation of anaerobic fermentation. Firstly, the mathematical model was established and then the energy, exergy, and economy analysis were evaluated. The results showed that the total exergy efficiency and the cost rate achieved 30.96% and 19.68$/h under design conditions. It was found that the total exergy efficiency increased as the increasing in the evaporating pressure of ORC and the basic ammonia solution concentration, but it decreased with the increase in the solar radiation intensity. When the input temperature and the current density of SOFC were increased, the total exergy efficiency was increased firstly then decreased and reached the maximum value of 439.2 and 481 kW at the SOFC input temperature of 628.6°C and the current density of 8286 A/m². Besides, the cost rate was increased with the increase of the power consumption of the main components. The Pareto frontier was obtained by using the non-dominated sorting whale optimization algorithm (NSWOA) which was employed to perform the multi-objective optimization, and the comprehensive decision-making method (TOPSIS) was used to get the optimal solution. The optimal solution showed that the total exergy efficiency and the cost rate could reach 39.56% and 14.23 $/h, respectively. Furthermore, in order to reduce optimization time and improve the accuracy of surrogate model, this study examined an improved artificial neural network (ANN) algorithm combining data-driven surrogate model and whale optimization algorithm (WOA) to replace the physical model. It was also found that the time using the developed surrogate model to obtain the optimal solution set spent only 5 minutes which is far less than the physical method which spent more than 40 hours under the same computer configuration.