Improvement of Dragonfly Algorithm with Chaos Theory

Providing an optimal solution that can produce issues of highcomplexity with the most appropriate time and probable outcome hasbeen a concern for many researchers. Hence, the methodologicalevolutionary algorithms, as an optimization method, have attracted agreat deal of interest in engineering applications, artificial intelligence,machine learning, and model recognition. Considering that in naturethe social behavior of many creatures is used for evolutionaryalgorithms, the behavior of the dragonfly bird has been of specialinterest and singularity. This insect forms small groups that travel wideareas to find food, which forms the phase of exploration, and in theextraction phase, they create large groups to go to different areas. Thedragonfly is powerful on exploring global areas for the reason wechose this algorithm to deal with lower convergence and localoptimum. Notice that chaos theory has non-repetitive sequel number,so it improves performance and gets better results.Dragonfly was evaluated with 13 benchmarks and differentconditions. In some functions, mean was reduced in minimum valuessuch as F1, and the result in F8 was 0.46 percent better than the basealgorithm. The results show that the proposed method has aremarkably better performance than other similar algorithms.