Dissertation > Excellent graduate degree dissertation topics show

Research and Application of the Simulated Annealing Particle Swarm Mixing Algorithm

Author: LinLingJuan
Tutor: LiuXiYu
School: Shandong Normal University
Course: Computer Software and Theory
Keywords: Particle Swarm Optimization Simulated annealing algorithm Gaussian mutation SA-DWPSO Image Classification
CLC: TP18
Type: Master's thesis
Year: 2010
Downloads: 224
Quote: 2
Read: Download Dissertation

Abstract

Since the 1980s, as an emerging field of swarm intelligence (Swarm Intelligence), causing the attention of many researchers, has become a hot frontier of artificial intelligence, as well as social, economic, biological, and other interdisciplinary. The artificial neural network, simulated annealing, genetic algorithms, particle swarm optimization algorithm and ant colony algorithm, developed by simulating certain natural phenomena and processes, and to provide new ideas and means of optimization theory. The 1995 particle swarm algorithm is simple and easy to implement, not much need to adjust the parameters, the convergence speed. Has been widely used in the objective function optimization, dynamic environment optimization, neural network training field, calculate the annual meeting and evolution in the IEEE (IEEE Annual Conference Of Evolutionary Computation, CEC) to become an independent research branches. The extension of the simulated annealing algorithm (Simulated Annealing, SA) as a local search algorithm, stochastic optimization methods to establish a simulated metal annealing mechanism by S.Kirkpatrick et al 1982. SA accepted a new model to become a global optimal algorithms and theoretical proof and verification of the actual application. It is precisely because of the existence of this advantage, it introduces the idea of ??the theory of combinatorial optimization. The algorithm in recent years has caused large-scale optimization design, extensive attention in the field of numerical analysis, complex layout. The rapid development of computer technology, multimedia technology and the Intemet technology lead to the emergence of a large number of images is a very important and challenging research topics: how to efficiently and quickly from a large-scale image database to retrieve the desired Figure elephant. Precisely in order to resolve the use of automatically obtain the image feature-based image retrieval technique, is retrieved from the image database a related image. In recent years, the study of this technology is very active, and have applications in many areas. The main content of this paper is to combine simulated annealing algorithm and particle swarm optimization algorithm, and around some of the key technology in the content-based image retrieval, some exploratory research. The content belongs to the image information retrieval and intelligent algorithms to optimize the focus of research in the field, has considerable theoretical significance and practical value, and provide a support platform for the design of the new ideas of pioneering intelligent classification and retrieval. The main work includes: 1. Proposed a dynamic adaptive particle swarm optimization the algorithms DAPSOPSO algorithm easy to fall into local optimum, premature convergence problem, many studies have focused on the improvement of the inertia weight w. Different inertia weight particle fulfill its responsibility, global optimization and local optimization simultaneously doing a good compromise between the guarantee algorithm global convergence and convergence rate. When the algorithm does not search the global best fitness value, or does not meet the optimal requirements, the inertia weight mutation strategy can be used. Can greater probability of a minor disturbance to the local search to achieve a big step long migration out of the local minimum area, but also due to a substantial disturbance. Proposed a \premature convergence, this paper proposes a \Dynamic small world particle swarm algorithm based on the introduction of hybridization and mutation mechanism, so as to reduce the computation time and avoid premature phenomenon. With particle swarm optimization algorithm and simulated annealing algorithm proposed hybrid algorithm of SA-DWPSO. Has proved in theory and particle swarm optimization algorithm and can not be guaranteed to converge to the optimal solution, even locally optimal solution. The simulated annealing algorithm has been proven according to the probability of convergence to a global optimal solution, so you can use the simulated annealing algorithm as the convergence of the PSO algorithm basis. Fusion through the local convergence of the simulated annealing particle swarm global convergence effectively overcome the premature convergence of particle swarm algorithm to speed up the convergence rate. Simulated annealing algorithm and improved particle swarm optimization both organic combination, cooperative search, can maintain the advantage of both the search, a good complementary. The hybrid algorithm is applied to image classification and retrieval based on image characteristics to achieve better classification results. Using VC NET 2008, SQL Server 2005 database system and MATLAB developed in the Windows XP platform complete. Respectively, for the retrieval of the image, classification, and optimization of the experimental analysis, design satisfactory results.

Related Dissertations

CLC: > Industrial Technology > Automation technology,computer technology > Automated basic theory > Artificial intelligence theory
© 2012 www.DissertationTopic.Net Mobile