Adaptation of improved hawk algorithm to multi objective optimization problems

In this study, Harris Hawks Optimization, which was developed with inspiration from hawks, have taken as an example. First stage; Preserving the structure of the HHO algorithm, an archive structure has been added to save and reuse Pareto optimal results. This archive is used to simulate hawk positions and solutions. The roulette wheel mechanism is used for the selection of solutions. The selection result is updated as the best result of the target. Thus, the Multi-Objective Harris Hawk Optimization algorithm has obtained. In the second stage; Multi Objective Harris Hawks Optimization algorithm based on Opposition Learning--based has been developed by adding the opposition learning mechanism on the MOHHO algorithm. It is aimed to increase the performance of the algorithm with the OppMOHHO algorithm. In order to better see the performance of the OppMOHHO algorithm, the unrestricted ZDT and DTLZ test function series in the literature have used. The thirteen test functions used have compared with OppMOHHO and four metaheuristic optimization algorithms found in the literature. For the statistical comparison of these algorithms, the Inverted Generational Distance, Generational Distance, Spacing, Spread and Maximum Spread metric have calculated. The results are given with tables and graphs. When the results are examined, for the OppMOHHO algorithm ZDT test function series; In the Inverted Generation Distance metric, OppMOHHO has quantitatively shown that it has high convergence behavior in the top three in the generation distance metric, in the top two in the ınterval metric, in the top three in the spread metric, and in the top two in the maximum spread metric. For DTLZ test function; The OppMOHHO algorithm for the ınverted generation distance metric, the top four for the generation distance metric, the top three for the ınterval metric, the OppMOHHO for the spread metric and the OppMOHHO for the maximum spread metric have quantitatively shown that it has high convergence behavior.