Dissertation > Excellent graduate degree dissertation topics show

Intelligent Group Optimization Algorithm PSO and Its Application in Several Types Models Optimization

Author: LiangHongSuo
Tutor: WangJianZhou
School: Lanzhou University
Course: Applied Mathematics
Keywords: Intelligent Optimization Algorithm Particle Swarm Optimization Uncertainty Grey Forecasting Model Grey Model Support Vector Machine The support vector machine multi- classification
CLC: TP18
Type: Master's thesis
Year: 2009
Downloads: 295
Quote: 3
Read: Download Dissertation

Abstract

Optimization technique is a mathematical basis for the application technology for solving various problems of the optimal solution. An important branch - the intelligent optimization algorithms, intelligent optimization algorithm is a through analog or explain some natural phenomenon or process developed from ordinary search algorithm is an iterative algorithm, a global, parallel and efficient optimize the performance, robustness, versatility and advantages. Uncertainty widely exists in the actual situation, has been the study of difficult problems. Uncertainty, it is proposed a lot of methods to deal with and they constitute an important branch of science, research and prediction uncertainty problem. The number of samples of these problems are usually very limited, and even a few, and most of the data sequence does not contain a clear quantitative relationship characteristics. This makes processing of these problems more difficult, many methods have achieved the better the effect, but these methods there can be areas of improvement. Optimization algorithm is introduced, with optimization algorithms to optimize the model, in order to improve the accuracy of the model. The main research achievements and contributions are as follows: 1) intelligent optimization algorithm applied to gray forecasting model GM (1,1) PSO and improved gray prediction model FGM (1,1). First, because the predictive ability of the model by means sequence of gray model, changing the the mean sequence parameters α will affect the predictive ability of the model, for the choice of the parameter α, generally use the default parameters or based on actual data and model predictions effects to set, with considerable arbitrariness is not a definite rule; once again, the gray model GM (1,1) requires a minimum of four data improved gray model FGM (1,1) requires a minimum of three data you can create a predictive model, required less number of data makes the data obtained by the model less, and this will inevitably affect the prediction of the model. In view of the above circumstances, the introduction of gray model through intelligent optimization algorithm PSO to optimize the model parameters α mean sequence optimization algorithm to search through the establishment of an appropriate fitness function, the search for a suitable parameter α will search parameter α application to improve the predictive ability of the model to the model. 2) intelligent optimization algorithm PSO applied to the support vector machine model, support vector machine model of the correct classification rate regression results by the impact of the punishment for coefficient of support vector machine, kernel function and kernel function parameters, different penalty coefficient have a great impact for the accuracy of the model and kernel function parameter, penalty coefficient and kernel function parameter selection is not a definite rule, often based on experience to select the value determined, so that the selected parameters for the different types of data are often not very suitable. Given this situation, through the introduction of intelligent optimization algorithm PSO support vector machine model, each depending on the data selected appropriate support the penalty coefficient vector machine and kernel function parameters, the establishment of an appropriate fitness function, using support vector optimization algorithm searches The machine suitable punishment for coefficient and the parameters of the kernel function, the search to punish coefficients and kernel function parameters applied to the model in order to improve the predictive ability of the model. 3) to test the effect of intelligent optimization algorithm PSO optimization: demonstrated, PSO optimization algorithm optimized gray forecasting model GM (1,1) predict the effect of the improved PSO optimization algorithm to optimize new gray prediction model FGM (1,1) improved. After PSO optimization algorithm optimized support vector machine multi-classification problem classification accuracy also improved significantly. Through empirical test the intelligent optimization algorithms PSO practicality.

Related Dissertations

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