Neural network-based allocation and self-improved firefly-based optimal sizing of fuel cells in distributed generation systems

The notion of Distributed Generation (DG) refers to the production of power at the level of consumption. Production of energy on-site, instead of offering it centrally, reduces the cost, internal dependencies, difficulties, inefficiencies, and risks that are related to transmission and distribution systems. In case DG is realized with fuel cells, several issues exist in respect to allocating and sizing of these fuel cells in the system. For solving those issues, dual stage intelligent technique is employed in this paper. First, the Neural Networks (NN) technique is adopted for determining the required location to place the fuel cells. Secondly, an enhanced version of Self Improved Fire-Fly (SIFF) algorithm is adopted for finding the optimal size of the fuel cells. The implemented technique is simulated in four IEEE benchmark test bus systems, and the respective performance analysis along with statistical analysis serves for validation purposes. The here proposed technique is compared with six other known algorithms, namely Particle Swarm Optimization (PSO), Firefly (FF) algorithm, Artificial Bee colony (ABC) algorithm, Improved Artificial Bee colony algorithm (IABC), Genetic Algorithm (GA) and Global Search Optimizer (GSO). The results obtained from the comparative analysis show the enhanced performance of the proposed mechanism