![]() | Only 14 pages are availabe for public view |
Abstract There are two categories of optimization techniques: exact and heuristic. Exact strategies guarantee the optimal solution will be found, and work well for many problems. However for complex problems or ones with a very large number of parameters, exact strategies may require very high computational costs. A large amount of real-world problems fall in this category of complex problems, and in order to solve them in a reasonable amount of time a different approach is needed For these problems, Meta-heuristic algorithms are considered as efficient tools to obtain near optimal solutions. In this thesis, we present BA as a new metaheuistic technique that has a great importance in the search area because of its accuracy and efficiency in solving optimization problems. We make a review of BA and its applications. It’s found that BA is an effective approach for solving a wide range of optimization problems and still needs more researches to improve and apply this method in other fields. The rest of the thesis includes three modified Bat algorithms; the first algorithm PBAF1 is concerned with enhancing the global search by modifying the generating phase in the original BA where the bats positions updating depends on the frequency which improve the convergence of the search and reduce the consumed time . The second algorithm PBAF2 is concerned with both global and local search where beside the previous modification in the global search, we modify the local search phase by inserting a control factor to its equation. The third modification PBA depends on controlling the convergence of the search by inserting a decreasing step weight factor to the position updating equation which control the convergence rate from the beginning to the end of the algorithm. These three modifications are evaluated using a number of benchmark problems on numeric optimization. The proposed modified algorithms have been implemented and tested on several state-of-the-art constrained and unconstrained benchmark optimization problems |