2019

Permanent URI for this collection

https://escholar.umt.edu.pk/handle/123456789/13364

Browse

Browsing 2019 by Issue Date

Filter results by year or month
Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Development of hybrid metaheuristic for global optimization
    (UMT Lahore, 2019-06-13) Javaid Ali
    Metaheuristics is a research area that delivers general purpose high quality optimization algorithms, proved effectual in dealing with complex global optimization problems. Success of metaheuristics greatly depends on their aptitude to establish equilibrium between their essential characters: exploration and exploitation. But the advent of No Free Lunch theorems by Wolpert and Macready established a general opinion that all algorithms perform equally when averaged over the whole function space and hence none of them can be claimed to be the best over the entire function space. For this reason, the basic algorithms require essential refinements and enhancements. The main goal of this thesis is twofold: to develop new effective hybrid metaheuristic strategies for solving selected global optimization problems and to analyze the performances of developed hybrid metaheuristics on mathematical benchmark functions and complex real world problems that can be modeled as global optimization problems. Generally, hybridization is carried out by integrating powerful components of different algorithms. The first hybrid metaheuristic proposed in this work is Controlled Showering Optimization (CSO) algorithm which is a combination of Artificial Showering Algorithm and frame based search mechanism. The second proposed hybrid algorithm is Cooperative Multi-Simplex algorithm (CMSA) that is based on collaborative search of multiple simplexes working under the iterations of a Non- Stagnated Nelder-Mead Simplex algorithm (NS-NMSA). The evolvement of the provably convergent variant NS-NMSA is also carried out in this work by identifying and coping the failures and stagnations of standard Nelder-Mead simplex algorithm. Multi-Simplex Imperialist Competitive Algorithm (MS-ICA) is the third hybrid metaheuristic which is designed by embedding NS-NMSA iterations in Imperialist Competitive Algorithm. The fourth hybrid metaheuristic designed in this continuation is obtained by integrating CMSA and Differential Evolution (DE) algorithm. In a specifically constructed computational framework, this hybrid algorithm in collaboration with Padé approximation is named as hybrid Evolutionary Padé Approximation (EPA) scheme.
  • Thumbnail Image
    Item
    Hybrid nature-inspired algorithms for engineering design optimization problems
    (UMT Lahore, 2019-11) Muhammad Luqman
    The focus of this dissertation is on the development of hybrid nature inspired metaheuristics for engineering design optimization problems. In this study, three nature inspired metaheuristics naming Artificial Showering Algorithm (ASHA), Artificial Bee Colony (ABC) algorithm and Differential Evolution (DE) have been considered for improvement and hybridization. We propose several improved as well as novel mixtures of the Nature Inspired Computational (NIC) methods, such as Targeted Showering Optimization (TSO), Radial ABC (RABC), hybrid of ABC and a modified ASHA (ABC-MASH) and Differential Targeted ABC (DTABC) algorithms. The structures and working principles of the proposed algorithms are discussed and analyzed in details. The performance of our proposed hybrid NIC algorithms has been investigated by statistical analysis of their results on nonlinear, unimodal, multi-modal, multi-objective, nonlinear systems in engineering and engineering design optimization problems. The analysis reveals that the proposed hybrid NIC algorithms overcome the deficiencies of individual algorithms and outperform several past hybrid methods on engineering design optimization problems. It has been established through computer simulations and non-parametric analysis of the results that our designed hybrid NIC algorithms are consistent in producing superior optimization results over the standard individual NIC algorithms as well as the past hybrid methods with respect to the exploration efficiency, speed of convergence and quality and quantity of the best and mean optimal solutions attained.