Adaptation and analysis of tree-seed algorithm for solving constrained and discrete optimization problems

Tree-seed algorithm (TSA) is a metaheuristic search algorithm proposed in 2015 to solve continuous unconstrained optimization problems. In this thesis, the studies have been carried out on the improvement and development of TSA, which produces high-quality results on small-sized continuous unconstrained optimization problems, in a way to provide solutions to continuous constrained, discrete binary and discrete integer optimization problems. A rule-based selection strategy which is in the literature has been integrated with TSA, and a constrained TSA (CTSA) has been proposed in order to solve continuous constrained optimization problems. Both numerical benchmark functions and engineering design problems were solved by CTSA, and it has been compared with the state-of-art algorithms in the literature. To solve discrete binary optimization problems, TSA is re-structured to work on binary search space, and logic-gate based TSA (LogicTSA) and similarity-based TSA (SimTSA) are proposed to generate new binary solutions. In order to obtain better results, these two approaches have been hybridized and a new approach, called SimLogicTSA, has been proposed, and more successful results have been obtained when compared to existing algorithms. In order to solve discrete integer optimization problems, TSA has been modified for working with permutation coded representation (DTSA). DTSA produces new solutions with neighborhood operators such as swap, shift, and symmetry. The quality of the solutions obtained by DTSA has been improved with the nearest neighbor heuristic and 2-opt local search approach. It has been shown that all proposed methods in this thesis content produce better quality or more competitive solutions than their counterparts in the literature.