الفهرس 
Title : Multi-objective Optimization Measures: A Comparative Analysis
SummaryOnly 14 pages are availabe for public view
 |  | from | 126 |  |  |
| | | from | 126 | | |
Abstract
The purpose of this thesis is to make a comparative analysis between a set of quality indices using a set of benchmark problems of different natures, type of constraints, number of decision variables, and different number of objective functions. Two famous optimization algorithms NSGA II [1], and SPEA 2 [2] well known are coded and used for solution of optimization problems. Some quality Indices are dependable and changeable according to the problem nature, number and type of constraints, number of decision variables, and number of objective function for optimization. Some quality Indices are eliminated and not significant in measuring the quality of optimization problems. Results indicate that Regular optimization algorithms are not able to handle constrained functions. Modification of the selection strategies for Genetic Algorithm can lead to set trade off solutions as close to Pareto over the entire objective space, and The modified NSGA II (M-NSGA II) is better than regular NSGA II or NSGA II with Penalty function approach with respect to both convergence and diversity. All Pareto front are evaluated in terms of 12 Quality Measures developed [1, 2, and 3]. These 12 Measures are then reduced to seven meaningful Quality indices. Benchmark problems with 2 and 3 objectives subject to equality and inequality constraints are employed