Open AccessEstimating PV models using multi-group salp swarm algorithm
Author(s) -
Mohammad Al-Shabi,
Chaouki Ghenaï,
Maâmar Bettayeb,
Fahad Ahmad,
Mamdouh El Haj Assad
Publication year - 2021
Publication title -
iaes international journal of artificial intelligence
Language(s) - English
Resource type - Journals
eISSN - 2252-8938
pISSN - 2089-4872
DOI - 10.11591/ijai.v10.i2.pp398-406
Subject(s) - algorithm , convergence (economics) , particle swarm optimization , swarm behaviour , group (periodic table) , computer science , mean squared error , photovoltaic system , mathematics , mathematical optimization , statistics , engineering , chemistry , organic chemistry , economics , economic growth , electrical engineering
In this paper, a multi-group salp swarm algorithm (MGSSA) is presented for estimating the photovoltaic (PV) solar cell models. The SSA is a metaheuristic technique that mimics the social behavior of the salp. The salps work in a group that follow a certain leader. The leader approaches the food source and the rest follows it, hence resulting in slow convergence of SSA toward the solution. For several groups, the searching mechanism is going to be improved. In this work, a recently developed algorithm based on several salp groups is implemented to estimate the single-, double-, triple-, Quadruple-, and Quintuple-diode models of a PV solar cell. Six versions of MGSSA algorithms are developed with different chain numbers; one, two, four, six, eight and half number of the salps. The results are compared to the regular particle swarm optimization (PSO) and some of its newly developed forms. The results show that MGSSA has a faster convergence rate, and shorter settling time than SSA. Similar to the inspired actual salp chain, the leader is the most important member in the chain; the rest has less significant effect on the algorithm. Therefore, it is highly recommended to increase the number of leaders and reduce the chain length. Increasing the number of leaders (number of groups) can reduce the root mean squared error (RMSE) and maximum absolute error (MAE) by 50% of its value.