Skip to Main Content
 

Global Search Box

 
 
 
 

ETD Abstract Container

Abstract Header

Bio-Inspired Evolutionary Algorithms for Multi-Objective Optimization Applied to Engineering Applications

DeBruyne, Sandra, DeBruyne

Abstract Details

2018, Doctor of Philosophy, University of Toledo, Engineering (Computer Science).
Although conventional mathematical methods for finding solutions to engineering applications have been around for many years, these methods are often quite computationally intensive when searching for an optimal solution. The idea of utilizing mathematically modeled animal hunting behaviors to find optimal solution sets to these types of problems is a relatively new field. These hunting behaviors, when incorporated into computerized optimization algorithms, have been shown in the literature to achieve results faster than other methods and return more accurate solutions with far fewer computational resources. The success of these bio-inspired algorithms to quickly search and intelligently predict the locations of the best solution set within the entire design space was the motivation for this research, the goal of which is to determine the applicability of this approach to engineering problem solving. Three research challenges present themselves when developing optimization algorithms for analyzing engineering problems. First, the number of problem objectives and constraints can range from two to hundreds. Second, the type of required interaction between the decision makers and the algorithm varies based on the type of engineering problem being optimized. And third, the algorithm’s computational time and the required amount of data storage for solutions can be become extremely large, which often prohibits successful optimization of large complex problems. Following extensive research on existing algorithms, it was determined that although there were many successful algorithms in the literature, they all exhibited shortcomings, and that they could be greatly improved. This determination lead to the development of two new bio-inspired multi-objective optimization algorithms. The new Grey Wolf Multi-Objective (GWMO) algorithm developed here utilizes the mathematical model for the choreographed hunting behaviors of a grey wolf pack to intelligently search for and quickly predict the most optimal evenly spaced solution sets for standard multi-objective problems sets. The new Harris’s Hawk Multi-Objective (HHMO) algorithm developed here utilizes the mathematical model for the cooperative hunting behaviors of hawks in surveying their environment from above to quickly direct the search for the location of the most optimal clustered solution sets for reference point multi-objective problem sets. This research had two objectives for determining its success at improving upon the multi-objective optimization algorithms that have been developed by others. The first objective was to analyze the effects of utilizing a-priori fitness methods over a-posteriori fitness methods within each of the GWMO and HHMO algorithm, since a-posteriori methods require large archival data storage spaces to maintain and analyze all possible solution sets. The GWMO and HHMO algorithms, with the combination of an a-priori method, evolutionary strategy and diversity method were shown to be successful at returning optimal solution sets without the need for this type of data storage. The second objective was to develop a novel flexible approach to algorithm development which could encompass both the wide and varied combinations of the number of objectives and the differing amounts of decision maker involvement required to optimize a variety of engineering applications. Both the GWMO and HHMO algorithms were successfully developed with this flexibility and showed a noticeable improvement over algorithms developed by others in optimizing a variety of engineering problems.
Devinder Kaur (Committee Chair)
Mansoor Alam (Committee Member)
Ivie Stein (Committee Member)
Ahmad Javaid (Committee Member)
Henry Ledgard (Committee Member)
Kevin Xu (Committee Member)
116 p.

Recommended Citations

Citations

  • DeBruyne, DeBruyne, S. (2018). Bio-Inspired Evolutionary Algorithms for Multi-Objective Optimization Applied to Engineering Applications [Doctoral dissertation, University of Toledo]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=toledo1542282067378143

    APA Style (7th edition)

  • DeBruyne, DeBruyne, Sandra. Bio-Inspired Evolutionary Algorithms for Multi-Objective Optimization Applied to Engineering Applications . 2018. University of Toledo, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=toledo1542282067378143.

    MLA Style (8th edition)

  • DeBruyne, DeBruyne, Sandra. "Bio-Inspired Evolutionary Algorithms for Multi-Objective Optimization Applied to Engineering Applications ." Doctoral dissertation, University of Toledo, 2018. http://rave.ohiolink.edu/etdc/view?acc_num=toledo1542282067378143

    Chicago Manual of Style (17th edition)