Browsing by Author "Dr. Daniel H. Loughlin, Member"

Filter results by typing the first few letters
Now showing 1 - 1 of 1
  • No Thumbnail Available
    Noninferior Surface Tracing Evolutionary Algorithm (NSTEA) for Multiobjective optimization
    (2000-08-16) Chetan, Srigiriraju Kishan; Dr. S. Ranji Ranjithan, Chair; Dr. E. Downey Brill Jr., Member; Dr. John W. Baugh Jr., Member; Dr. Daniel H. Loughlin, Member
    Evolutionary algorithms are becoming increasingly valuable in solving large-scale, realistic engineering problems. Most of these problems deal with sufficiently complex issues that typically conflict with each other, thus requiring multi objective (MO) analyses to assist in identifying compromise solutions. The focus of this paper is to develop and test a new multi objective evolutionary algorithm (MOEA). The new procedure, Noninferior Surface Tracing Evolutionary Algorithm (NSTEA), builds upon two fundamental concepts that are established in the mathematical programming literature for MO analysis. Implicit implementation of Pareto optimality and beneficial seeding of initial population are instrumental in the improved performance. NSTEA was evaluated by solving a suite of test problems reported in the MOEA literature. Performance with respect to accuracy, coverage, and spread of noninferior solutions generated by NSTEA is evaluated and compared with those of solutions generated by four other MOEAs that are widely accepted. Also, in some cases, comparisons are made with noninferior sets generated using mathematical programming techniques. Overall, NSTEA performs relatively better than the other MOEAs when tested on these problems. Application and performance evaluation of NSTEA in solving a real-world MO engineering optimization problem was also conducted. In comparison to published mathematical programming-based noninferior solutions, the NSTEA solutions performed well. In summary, this paper contributes to the MOEA literature by presenting NSTEA as a good alternative evolutionary algorithm-based multi objective method that is relatively simple to implement and to incorporate into existing implementations of evolutionary algorithm-based optimization procedures.