No description available.

No description available.

Call number: LD2668 .T4 1960 M55

published_or_final_version / Mathematics / Master / Master of Philosophy

by Anthony Yiu-Cheung Tang. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1994. / Includes bibliographical references (leaves 145-152). / ACKNOWLEDGMENTS --- p.i / ABSTRACT --- p.ii / LIST OF FIGURES --- p.vii / LIST OF TABLES --- p.ix / Chapter 1. --- INTRODUCTION --- p.1 / Chapter 1.1 --- Introducing evolutionary computation --- p.2 / Chapter 1.2 --- Adaptation and learning --- p.7 / Chapter 1.3 --- Comparing the efficency of evolutionary computation and sequential computation --- p.8 / Chapter 1.4 --- The place of evolutionary computation in computer science --- p.9 / Chapter 1.4.1 --- Mathematical foundation --- p.9 / Chapter 1.4.2 --- Scalability --- p.10 / Chapter 1.4.3 --- Parallelism --- p.11 / Chapter 1.5 --- Enhancing genetic search by local search --- p.11 / Chapter 1.6 --- Thesis Overview --- p.12 / Chapter 2. --- A REVIEW OF GENETIC ALGORITHMS --- p.14 / Chapter 2.1 --- Introduction --- p.14 / Chapter 2.2 --- The canonical genetic algorithm --- p.14 / Chapter 2.3 --- Optimal allocation of trials and schemata analysis --- p.17 / Chapter 2.4 --- Applications --- p.23 / Chapter 2.4.1 --- Function optimizations --- p.23 / Chapter 2.4.2 --- Machine Learning --- p.24 / Chapter 2.4.3 --- Combinatorial optimizations --- p.25 / Chapter 2.5 --- Criticisms --- p.25 / Chapter 2.5.1 --- Parameter settings --- p.25 / Chapter 2.5.2 --- Convergence and divergence --- p.26 / Chapter 2.5.3 --- Genetic algorithms for function optimizations --- p.27 / Chapter 2.5.4 --- The role of crossover and build blocks --- p.28 / Chapter 2.6 --- Future directions --- p.29 / Chapter 2.6.1 --- Is the schemata theorem wrong ？ --- p.29 / Chapter 2.6.2 --- Artificial life --- p.29 / Chapter 2.6.3 --- Parallel genetic algorithms --- p.31 / Chapter 2.6.4 --- Non-binary alphabets --- p.31 / Chapter 2.6.5 --- Investigations on problems that are hard for GA --- p.33 / Chapter 3. --- THE GENERAL STRUCTURE OF TOLKIEN --- p.34 / Chapter 3.1 --- Introduction --- p.34 / Chapter 3.2 --- Class Description --- p.39 / Chapter 3.2.1 --- Collection classes --- p.39 / Chapter 3.2.2 --- Vector classes --- p.39 / Chapter 3.2.3 --- GA-related classes --- p.40 / Chapter 3.2.4 --- Utility classes --- p.42 / Chapter 3.3 --- The TOLKIEN Genetic Algorithm --- p.43 / Chapter 3.3.1 --- Binary and Gray Code Representations --- p.44 / Chapter 3.3.2 --- Crossover Operators --- p.44 / Chapter 3.3.3 --- Haploids and Diploids --- p.47 / Chapter 3.3.4 --- Population --- p.50 / Chapter 3.3.5 --- Selection scheme --- p.50 / Chapter 3.3.6 --- Scaling scheme...: --- p.51 / Chapter 3.4 --- The TOLKIEN Classifier System --- p.52 / Chapter 3.4.1 --- Classifiers --- p.52 / Chapter 3.4.2 --- Messages and Message Lists --- p.53 / Chapter 3.4.3 --- Producing New Messages --- p.55 / Chapter 3.4.4 --- The Bucket Brigade Algorithm --- p.55 / Chapter 3.5 --- Where to obtain TOLKIEN --- p.56 / Chapter 4. --- ILLUSTRATING THE CAPABILITIES OF TOLKIEN --- p.57 / Chapter 4.1 --- de Jong's Test Bed : Function Optimization using GA --- p.57 / Chapter 4.2 --- Royal road function experiments --- p.63 / Chapter 4.2.1 --- RRMF --- p.64 / Chapter 4.2.2 --- RRJH --- p.65 / Chapter 4.2.3 --- Testing royal road functions using TOLKIEN --- p.68 / Chapter 4.2.4 --- Results --- p.71 / Chapter 4.2.5 --- Adding hillclimbing algorithm to solve royal road functions --- p.72 / Chapter 4.2.6 --- Discussions --- p.73 / Chapter 4.3 --- A classifier system to learn a multiplexer --- p.74 / Chapter 4.4 --- A classifier system maze traveller --- p.83 / Chapter 4.4.1 --- Framework of the Animat --- p.84 / Chapter 4.4.2 --- Constructing the maze navigation classifier system --- p.85 / Chapter 4.4.3 --- Results --- p.86 / Chapter 4.5 --- Future Enhancements on TOLKIEN --- p.88 / Chapter 4.6 --- Chapter Summary --- p.88 / Chapter 5. --- SOLVING TSP USING GENETIC ALGORITHMS --- p.89 / Chapter 5.1 --- Introduction --- p.89 / Chapter 5.2 --- Recombination operators for TSP --- p.91 / Chapter 5.2.1 --- PMX Crossover --- p.91 / Chapter 5.2.2 --- Order Crossover --- p.92 / Chapter 5.2.3 --- Edge Recombination operator --- p.93 / Chapter 5.3 --- Simulated Annealing --- p.95 / Chapter 5.4 --- Simulation Comparisons --- p.96 / Chapter 5.4.1 --- The Test Bed --- p.96 / Chapter 5.4.2 --- The Experimental Setup --- p.97 / Chapter 5.4.3 --- Results --- p.97 / Chapter 5.4.4 --- Discussions --- p.100 / Chapter 6. --- AN IMPROVED EDGE RECOMBINATION OPERATOR FOR TSP --- p.101 / Chapter 6.1 --- EDGENN ： a new edge recombination operator --- p.102 / Chapter 6.2 --- Experimental results --- p.104 / Chapter 6.2.1 --- Comparing EdgeNN and Edge-2 --- p.104 / Chapter 6.2.2 --- Comparing EdgeNN and Edge-3 --- p.106 / Chapter 6.3 --- Further improvement : a heuristic genetic algorithm using EdgeNN --- p.106 / Chapter 6.4 --- Discussion --- p.108 / Chapter 7. --- CONCLUSIONS --- p.111 / Chapter 7.1 --- Evaluation on TOLKIEN --- p.111 / Chapter 7.2 --- EdgeNN as a useful recombination operator for solving TSP --- p.112 / Chapter 7.3 --- Genetic algorithm and hillclimbing --- p.112 / EPILOGUE --- p.113 / APPENDIX : PROGRAM LISTINGS --- p.114 / Function optimizations --- p.114 / Maze Navigator --- p.122 / Multiplexer --- p.135 / Royal road functions --- p.141 / BIBLIOGRAPHY --- p.145 / INDEX --- p.153

by Lee, Ka-wai. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1995. / Includes bibliographical references (leaves 61-67). / Chapter 1 --- Introduction --- p.6 / Chapter 1.1 --- Traveling Salesman Problem --- p.6 / Chapter 1.2 --- Genetic Algorithms --- p.8 / Chapter 1.3 --- Solving TSP using Genetic Algorithms --- p.10 / Chapter 1.4 --- Outline of Work --- p.12 / Chapter Part I --- Algorithm Development --- p.14 / Chapter 2 --- A Local DP Crossover Operator 一 LDPX --- p.15 / Chapter 2.1 --- Review of DP for Solving TSP --- p.15 / Chapter 2.2 --- On the Original LDPX --- p.18 / Chapter 2.2.1 --- Gene Representation --- p.18 / Chapter 2.2.2 --- The Original Crossover Procedure --- p.19 / Chapter 2.3 --- Analysis --- p.21 / Chapter 2.3.1 --- Ring TSP --- p.21 / Chapter 2.3.2 --- Computational Results of Solving Ring TSP and Other TSP using LDPX --- p.22 / Chapter 2.4 --- Augmentation of the Gene Set Representation --- p.24 / Chapter 2.5 --- Enhancement of Crossover Procedure --- p.25 / Chapter 2.6 --- Computational Comparison of the new proposed LDPX with the orig- inal LDPX --- p.26 / Chapter 2.7 --- SPIR ´ؤ An Operator for Single Parent Improved Reproduction --- p.26 / Chapter 3 --- A New TSP Solver --- p.29 / Chapter 4 --- Performance Analysis of the TSP Solver --- p.33 / Chapter 4.1 --- Computational results --- p.34 / Chapter 4.2 --- "Comparison between SPIR/LDPX, PMX and ER" --- p.35 / Chapter 4.3 --- Convergence Test of SPIR/LDPX --- p.37 / Chapter Part II --- Application --- p.43 / Chapter 5 --- Flowshop Scheduling Problem --- p.44 / Chapter 5.1 --- Brief Review of the Flowshop Scheduling Problem --- p.44 / Chapter 5.2 --- Flowshop Scheduling with travel times between machines --- p.45 / Chapter 6 --- A New Approach to Solve FSTTBM --- p.47 / Chapter 7 --- Computational Results of the New Algorithm for CPFSTTBM --- p.53 / Chapter 7.1 --- Comparison with Global Optimum --- p.54 / Chapter 7.2 --- The Algorithm of SPIRIT --- p.55 / Chapter 7.3 --- Comparison with SPIRIT --- p.57 / Chapter 8 --- Conclusion --- p.59 / Bibliography --- p.61 / Chapter A --- Random CPFSTTBM problem Generation Algorithm --- p.68

We prove the existence of a one-parameter family of solutions of the porous medium equation, a nonlinear heat equation. In our work, with space dimension 3, the interface is a half line whose end point advances at constant speed. We prove, by using maximum principle, that the solutions are stable under a suitable class of perturbations. We discuss the relevance of our solutions, when restricted to two dimensions, to gravity driven flows of thin films. Here we extend the results of J. Iaia and S. Betelu in the paper "Solutions of the porous medium equation with degenerate interfaces" to a higher dimension.

Graduation date: 1997

Extra High Voltage (EHV) transmission lines are designed to transfer large amount of power from one location to another. The length exposed to the environment is a major reason for occurrence of faults on the lines. A fault on a high voltage transmission line affects the stability of the overall power system, which sometimes leads to permanent damage of the equipment. Relays are developed and installed to protect the lines. The transmission line protection relays, in the industry, are based on the fundamental frequency components of the voltages and currents. These relays need at least one fundamental frequency cycle for performing the protection operation. Voltage and current traveling waves are generated when a fault occurs on the transmission line. The velocity of propagation of traveling waves is finite and the level of the waves decreases with increase in the distance traveled. Information about the fault can be obtained by analyzing the traveling waves. A few traveling wave techniques, which are based on analog signal processing, to protect transmission lines have been proposed in the past. Two digital techniques, which use traveling waves for protecting EHV transmission lines, are proposed in this thesis. The traveling waves are extracted from the modal voltages and currents at the terminals of the transmission line. The techniques identify and locate the fault by using the information contained in the waves. A power system was modeled in the Electromagnetic Transient Direct Current Analysis (EMTDC) and several cases were created by varying different parameters related to the fault, fault type, fault location, fault resistance and fault inception angle. The techniques were implemented in hardware and their performance was tested on data, generated from the EMTDC simulations. Some cases are discussed in the thesis. The performance of the digital techniques for protecting EHV transmission lines using traveling waves was confirmed to be satisfactory. The proposed techniques provide protection at speed and discriminate well between internal and external faults.

Extra High Voltage (EHV) transmission lines are designed to transfer large amount of power from one location to another. The length exposed to the environment is a major reason for occurrence of faults on the lines. A fault on a high voltage transmission line affects the stability of the overall power system, which sometimes leads to permanent damage of the equipment. Relays are developed and installed to protect the lines. The transmission line protection relays, in the industry, are based on the fundamental frequency components of the voltages and currents. These relays need at least one fundamental frequency cycle for performing the protection operation. Voltage and current traveling waves are generated when a fault occurs on the transmission line. The velocity of propagation of traveling waves is finite and the level of the waves decreases with increase in the distance traveled. Information about the fault can be obtained by analyzing the traveling waves. A few traveling wave techniques, which are based on analog signal processing, to protect transmission lines have been proposed in the past. Two digital techniques, which use traveling waves for protecting EHV transmission lines, are proposed in this thesis. The traveling waves are extracted from the modal voltages and currents at the terminals of the transmission line. The techniques identify and locate the fault by using the information contained in the waves. A power system was modeled in the Electromagnetic Transient Direct Current Analysis (EMTDC) and several cases were created by varying different parameters related to the fault, fault type, fault location, fault resistance and fault inception angle. The techniques were implemented in hardware and their performance was tested on data, generated from the EMTDC simulations. Some cases are discussed in the thesis. The performance of the digital techniques for protecting EHV transmission lines using traveling waves was confirmed to be satisfactory. The proposed techniques provide protection at speed and discriminate well between internal and external faults.