英语论文网

This research presents contributions in three fundamental areas of metaheuristics for the optimization of engineering problems: design and analysis of metaheuristic algorithms, improvements of existing metaheuristics and methods for parameter control, and hybrid metaheuristics. The primary objectives of this thesis can be summarized as follows:
?To design and analyze metaheuristic algorithms
?To develop new algorithms to extend and improve the existing GA and PSO algorithms and methods for parameter control
?To perform enhancements to metaheuristics algorithms through the hybridization of the algorithms, leading to new memetic algorithms
?To investigate the application of the proposed algorithms to the task of optimization of engineering problems

1.7 论文概述——1.7 Dissertation Overview

We now give a summary of the following chapters of this thesis:

In Chapter 2, a popular metaheuristic algorithm; Genetic Algorithm is analyzed and implemented on a set of high-level synthesis (HLS) benchmarks. The goal is to minimize area and maximize the throughput. Results from the algorithms are observed and analyzed to suggest more efficient algorithms.

Chapter 3 presents two new adaptive GAs to automate the parameter selection and operator-probability adaptation. A detailed comparison between a standard GA and the adaptive algorithms will be presented. In addition, techniques to improve the control over the population diversity are investigated. The guided GA (GGA) that uses a diversity measure to alternate between exploring and exploiting behavior of the GA is designed. In addition, two other enhancements have been proposed for GA. These include the use of a set of genetic operators that perform directional mutation and organizing selection tournaments based on the genotype vicinity.

In Chapter 4, a novel approach employing PSO in conjunction with list scheduling to create a new hybrid algorithm is introduced. The local and global search heuristics in PSO are iteratively adjusted making it an effective technique for exploring the discrete solution space for an optimal solution. To avoid premature convergence of the PSO algorithm, the PSO algorithm is modified to include a new diversity measure to control the swarm in phases of attraction and repulsion (ARPSO). Additionally, several enhancements to PSO based on diversity, efficient initialization using different distributions and low-discrepancy sequences is proposed. The performance of PSO and GA is extensively compared and the relative merits of both approaches are given.

Next, memetic algorithm that combines a metaheuristic algorithm to perform exploration and the local search to perform exploitation is introduced in Chapter 5. Two new MAs were designed: the first combines the adaptive GA with local search and was tested on the HLS benchmarks. Additionally, the balance between genetic search and local search was investigated based on four major factors: early termination of local search, restriction on the number of solutions to which local search is applied and the frequency of applying local search and adapting the n本论文由英语论文网提供整理，提供论文代写，英语论文代写，代写论文，代写英语论文，代写留学生论文，代写英文论文，留学生论文代写相关核心关键词搜索。