الفهرس 
Chapter 1 : IntroductionOnly 14 pages are availabe for public view
 |  | from | 191 |  |  |
| | | from | 191 | | |
Abstract
Optimization is a systematic mathematical process of identifying the best optimal solution among a set of solutions for the same problem by choosing the most appropriate values for the optimization problem decision variables. This process is done automatically by the optimization algorithm which tunes the problem parameters in a specified range of values. There are many types of optimization algorithms.
Among these types, metaheuristics have been considered as globally optimal solution-oriented methods as these algorithms provide a global optimal solution for the problem in hand and prevent the appearance of problems faced in other types of optimization algorithms such as heuristic algorithms.
Trapping into local optima problems, the unbalance between the exploration and exploitation abilities, the slow convergence to the optimal solutions, and the inconsistency in the obtained solutions are the most common defects handled by the metaheuristic algorithms.
Besides the high quality of solutions obtained by the metaheuristics, a set of merits is owned by these methods. Among these merits are the inspiration simplicity, in which we can simply inspire new algorithms from different phenomena or natural concepts in different subject areas, and the flexibility, where we can use metaheuristics to solve diverse types of problems by formulating the problem as an optimization problem. Benefiting from the previously mentioned merits and others, the researchers recently concentrated their works on studying the existing metaheuristic algorithms, enhancing their performance, and employing them in solving different classes of problems in many subject areas.
Based on that, this thesis introduces the following:-
1) A comprehensive review about optimization algorithms and the optimization problems. Specifically, this review presents different taxonomies of the optimization algorithms, the types of optimization problems, the development process of the new optimization algorithms, as well as the novelty claims of novel based optimization algorithm. one of the main points of this review is to pay more attention to the open issues and challenges in the metaheuristic field for redirecting researchers to future trends and active points related to this field.
2) Integrating two of the most popular enhancement strategies to raise the performance of the recently published Exponential Distribution Optimizer (EDO) and employing the enhanced version (mEDO) in solving Global and Engineering design and combinatorial optimization problems.
3) Boosting the performance and applicability of the White Sharks Optimizer(WSO) by emerging two of the most popular enhancement mechanisms to propose an enhanced White Shark Optimizer (EWSO) and use it for solving global and engineering design, as well as two combinatorial optimization problems.
The main findings are as follows:
• Eventually, this thesis introduces a comprehensive review about the optimization algorithms and the optimization problems, the development process of the optimization algorithm, the evaluation environment of the proposed algorithms, in addition to redirect the new researchers to the future trends and active points in the area of optimization.
• The first proposed EWSO algorithm has achieved the first rank compared to the other seven competing algorithms in solving the CEC’2022 benchmark test functions, and eight engineering design problems in addition to achieving very promising results in handling two combinatorial quadratic assignment(QAP) and Bin packing problems(BPP), also achieving the results in compared to other algorithms tackled the same instances of QAP and BPP.
• The second proposed mEDO algorithm has achieved the first rank compared to the other seven competing algorithms in solving the CEC’2020 benchmark test functions with dimension 20, and the fist rank in comparing the proposed mEDO algorithm with the recent and winners’ algorithms in dimension 50 and 100. The proposed mEDO algorithm has achieved a promising result in solving eight engineering design problems in addition to achieving very promising results in handling the combinatorial quadratic assignment problems.
Conclusions
• In this thesis, a comprehensive study of metaheuristics algorithms is introduced involving definitions of the concept of optimization, studying the appearance of metaheuristic terms, introducing an explanation of the features of the MAs more than other techniques, different taxonomies of MAs according to different aspects such as inspiration source, number of search agents, population updating mechanisms, and number of parameters. in addition to studying the metrics used in the Performance evaluation of the algorithm.
• A significant effort is paid to clarify the optimization problem in detail, concentrating on different classification techniques, and the study reviews the use of metaheuristics in different application areas such as engineering design problems, NP-hard problems, medical science, and robotics. Finally, we introduce some of the issues that exist in the MAs literature and the future directions of this important field.
• from the applicability perspective, we have developed a new version of the White Shark Optimizer (WSO) called EWSO, which combines two enhancement strategies; Orthogonal Learning (OL) and Enhanced Solution Quality (ESQ) in order to boost the performance of the original WSO algorithm and beat the shortcomings of the original WSO, which are stuck in local optima problems and poor balance between exploitation and exploration, particularly when solving high-dimension problems.
• To confirm the quality and performance of the proposed EWSO algorithm, we have evaluated its performance over twelve functions from the IEEE CEC’2022 benchmark and three constrained engineering design optimizations problems in addition to two combinatorial problems.
• The proposed EWSO has been compared with seven other metaheuristic algorithms, including the original (i.e., WSO). The results obtained by EWSO demonstrate its superior ability to locate the optimal region, achieve a better trade-off between exploring and exploiting mechanisms, and converge faster to a nearly optimal solution than other algorithms.
• Furthermore, statistical analysis reveals that EWSO outperforms other algorithms. Overall, the EWSO is considered a useful optimization tool to solve other optimization problems.
• Also, this study has presented a new version of the Exponential Distribution Optimizer (EDO) algorithm called mEDO, which combines two enhancement strategies; Orthogonal Learning (OL) and Local Escaping Operator (LEO) in order to improve the performance of the original EDO algorithm.
• To validate the quality and performance of the proposed mEDO algorithm, ten functions from the IEEE CEC’2020 benchmark and eight constrained engineering design optimization problems have been used.
• The proposed mEDO has been compared with seven other metaheuristic algorithms, including the original algorithm (i.e., EDO). The results obtained by mEDO demonstrate its superior ability to locate the optimal region, achieve a better trade-off between exploring and exploiting mechanisms, and converge faster to a nearly
optimal solution than other algorithms. Furthermore, statistical analysis reveals that mEDO outperforms other algorithms.
• Finally, this thesis presents the active points in the optimization field to redirect the newcomer’s researchers in continuing the journey of introducing the concept of optimization in many other application
areas.