Self-adaptive salp swarm algorithm for engineering optimization problems

• Published in: Applied Mathematical Modelling Vol. 89; pp. 188 - 207
• Main Authors: Salgotra, Rohit, Singh, Urvinder, Singh, Supreet, Singh, Gurdeep, Mittal, Nitin
• Format: Journal Article
• Language: English
• Published: New York Elsevier Inc 01.01.2021; Elsevier BV
• Subjects: Adaptive algorithms; Adaptive control; Algorithms; Cognitive radio; Cognitive radio system; Computation intelligence; Convergence; Exploitation; Machine learning; Optimization; Parameter modification; Salp swarm algorithm; Self-adaptive parameters; Statistical tests; Salp swarm algorithm; Self-adaptive parameters; Computation intelligence; Cognitive radio system; Optimization
• ISSN: 0307-904X; 1088-8691; 0307-904X
• DOI: 10.1016/j.apm.2020.08.014

•Division of iteration and adaptive parameters to balance exploration and exploitation.•Enhanced exploitation phase based on the hybrid concepts of grey wolf optimization and cuckoo search algorithm.•Performance evaluation for variable population and dimension using numerical benchmarks.•Linearly decreasing population size to reduce the computational burden of salp swarm algorithm.•Real world transmission parameter optimization of cognitive radio system. Salp swarm algorithm is a recent introduction in the field of swarm intelligent algorithms and has proved its worth over various research domains. Though it is a competitive algorithm but it has been found that salp swarm algorithm suffers from various problems including poor exploitation, slow convergence and unbalanced exploration and exploitation operation. In present work, four major modifications have been added to salp swarm algorithm in order to make it self-adaptive and the proposed algorithm has been named as adaptive salp swarm algorithm. The modifications include division of generations and logarithmic adaptive parameters to control the extent of exploration and exploitation, enhanced exploitation phase to improve the local search and linearly decreasing population adaptation to reduce the total number of function evaluations. The performance of the proposed algorithm is tested on benchmark problems and further applied for optimization of transmission parameters in cognitive radio system. From the experimental results, it has been found that the proposed adaptive salp swarm algorithm is highly competitive and provides better results when compared with bat algorithm, grey wolf optimization, teacher learning based algorithm, dragonfly algorithm and others. Convergence profiles and statistical tests further validate the results.