Thesis (M.Eng.)(Mechanical)--University of Pretoria, 2005. / Summaries in English and Afrikaans. Includes bibliographical references.

Thesis (Ph. D.)--University of Washington, 2005. / Vita. Includes bibliographical references (p. 135-139).

Thesis (M.S.)--Rochester Institute of Technology, 2007. / Typescript. Includes bibliographical references (leaves 88-91).

Object or part recognition is of major interest in industrial environments. Current methods implement expensive camera based solutions. There is a need for a cost effective alternative to be developed. One of the proposed methods is to overcome the hardware, camera, problem by implementing a software solution. Artificial Neural Networks (ANN) are to be used as the underlying intelligent software as they have high tolerance for noise and have the ability to generalize. A colleague has implemented a basic ANN based system comprising of an ANN and three cost effective laser distance sensors. However, the system is only able to identify 3 different parts and needed hard coding changes made by trial and error. This is not practical for industrial use in a production environment where there are a large quantity of different parts to be identified that change relatively regularly. The ability to easily train more parts is required. Difficulties associated with traditional mathematically guided training methods are discussed, which leads to the development of a Genetic Algorithm (GA) based evolutionary training method that overcomes these difficulties and makes accurate part recognition possible. An ANN hybridised with GA training is introduced and a general solution encoding scheme which is used to encode the required ANN connection weights. Experimental tests were performed in order to determine the ideal GA performance and control parameters as studies have indicated that different GA control parameters can lead to large differences in training accuracy. After performing these tests, the training accuracy was analyzed by investigation into GA performance as well as hardware based part recognition performance. This analysis identified the ideal GA control parameters when training an ANN for the purpose of part recognition and showed that the ANN generally trained well and could generalize well on data not presented to it during training.

No description available.

Genetic Algorithms (GAs) are powerful tools to solve large scale design optimization problems. The research interests in GAs lie in both its theory and application. On one hand, various modifications have been made on early GAs to allow them to solve problems faster, more accurately and more reliably. On the other hand, GA is used to solve complicated design optimization problems in different applications. The study in this thesis is both theoretical and applied in nature. On the theoretical side, an improved GA�Evolution Direction Guided GA (EDG-GA) is proposed based on the analysis of Schema Theory and Building Block Hypothesis. In addition, a method is developed to study the structure of GA solution space by characterizing interactions between genes. This method is further used to determine crossover points for selective crossover. On the application side, GA is applied to generate optimal tolerance assignment plans for a series of manufacturing processes. It is shown that the optimal tolerance assignment plan achieved by GA is better than that achieved by other optimization methods such as sensitivity analysis, given comparable computation time.

The purpose of this study is to discuss the fitness of using PHMGA (Parallel Multilevel Hybrid Genetic Algorithm), which is a fast and efficient method, in the geometrically nonlinear structural optimization. Parallel genetic algorithms can solve the problem of traditional sequential genetic algorithms, such as premature convergence, large number of function evaluations, and a difficulty in setting parameters. By using several concurrent sub-population, parallel genetic algorithms can avoid premature convergence resulting from the single genetic searching environment of sequential genetic algorithms. It is useful to speed up the operation rate of joining timely multilevel optimization with parallel genetic algorithms. Because multilevel optimization can resolve one problem into several smaller subproblems, each subproblem is independent and not interference with one another. Then the subsystem of each level can be connected by sensitivity analysis. So we can solve the entire problem. Because each subproblem contains less variables and constrains, it can achieve the faster converge rate of the entire optimization. PHMGA integrates advantages of two methods including the parallel genetic algorithms and the multilevel optimization. In this study, PHMGA is adopted to solve several design optimization problems for nonlinear geometrically trusses on the parallel computer IBM SP2. The use of PHMGA helps reduce execution time because of integrating a multilevel optimization and a parallel technique. PHMGA helps speed up the searching efficiency in solving structural optimization problems of nonlinear truss. It is hoped that this study will demonstrate PHMGA is an efficient and powerful tool in solving large geometrically nonlinear structural optimization problems.

Genetic Algorithms (GAs) are search algorithms inspired by genetics and natural selection, and have been used to solve difficult problems in many disciplines, including modelling, control systems and automation. GAs are generally able to find good solutions in reasonable time, however as they are applied to larger and harder problems they are very demanding in terms of computation time and memory. The Internet is the most powerful parallel and distributed computation environment in the world, and the idle cycles and memories of computers on the Internet have been increasingly recognized as a huge untapped source of computation power. By combining Internet computing and GAs, this dissertation provides a framework for Internet-based parallel and distributed GAs that gives scientists and engineers an easy and affordable way to solve hard real world problems. Developing parallel computation applications on the Internet is quite unlike developing applications in traditional parallel computation environments, such as multiprocessor systems and clusters. This is because the Internet is different in many respects, such as communication overhead, heterogeneity and volatility. To develop an Internet-based GA, we need to understand the implication of these differences. For this purpose, a convergence model for heterogenous and volatile networks is presented and used in experiments that study GA performance and robustness in Internet-like scenarios. The main outcome of this research is an Internet-based distributed GA framework called G2DGA. G2DGA is an island model distributed GA, which can provide support for big populations needed to solve many real world problems. G2DGA uses a novel hybrid peer-to-peer (P2P) design with island node activity coordinated by supervisor nodes that offer a global overview of the GA search state. Compared to client/server approaches, the P2P architecture improves scalability and fault tolerance by allowing direct communication between the islands and avoiding single-point-of-failure situations. One of the defining characteristics of Internet computing is the dynamics and volatility of the environment, and a parallel and distributed GA that does not adapt to its environment cannot use the available resources efficiently. Two novel adaptive methods are investigated. The first method is migration topology adaptation, which uses clustering on elite individuals from each island to rebuild the migration topology. Experiments with the migration topology adapter show that it gives G2DGA better performance than a GA with static migration topology of a similar or larger connectivity level. The second method is population size adaptation, which automatically finds the number of islands and island population sizes needed to solve a given problem efficiently. Experiments on the population size adapter show that it is robust, and compares favourably with the traditional trial-and-error approach in terms of computational effort and solution quality. The scalability and robustness of G2DGA has been extensively tested in network scenarios of varying volatility and heterogeneity. Experiments with up to 60 computers were conducted in computer laboratories, while more complex network scenarios have been studied in an Internet simulator. In the experiments, G2DGA consistently performs as well as, and usually significantly better than, static distributed GAs and the difference grows larger with increased network instability. The results show that G2DGA, by continuously adjusting the migration policy and the population size, can detect and make efficient use of idle cycles donated over volatile Internet connections. To demonstrate that G2DGA can be used to implement and solve real world problems, a challenging application in VLSI design was developed and used in the testing of the framework. The application is a multi-layer floorplanner, which uses a novel GA representation and operators based on a slicing structure approach. Its packing quality compares favourably with other multi-layer floorplanners found in the literature. Internet-based distributed GA research is exciting and important since it enables GAs to be applied to problem areas where resource limitations make traditional approaches unworkable. G2DGA provides a scalable and robust Internet-based distributed GA framework that can serve as a foundation for future work in the field.

No description available.

No description available.