Font Size: a A A

An Intelligent Exam-paper Generatingmethod Based On Improved Genetic Algorithm

Posted on:2010-06-05Degree:MasterType:Thesis
Country:ChinaCandidate:Y P JinFull Text:PDF
GTID:2178360272980292Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
In recent years, people were very concerned about intelligent optimization algorithm, such as artificial neural networks, genetic algorithms.to address the complex issues provided a new approach, and in many areas of success. The problem of proposed papers was a certain condition of the multi-objective parameters optimization problems, in the light of the shortcomings of algorithm for the traditional proposed paper with a slow speed, the success rate was lower and paper quality was not high, through detailed analysis of the papers of the constraints such as knowledge, the degree of difficulty, discrimination, created a mathematical model of intelligent test paper, the use of improved genetic algorithm to achieve the intelligence proposed paper.The improved genetic algorithms using real-coded. numbers of the questions of composed papers directly, for gene mapping, using real-coded to avoid in the encoding and decoding process brought about by the loss of time and improved the efficiency of operations; improved methods to generate the initial group, making the success rate of composed papers and quality significantly improved; genetic improvement of operations, and improved cross-adaptive and mutation, the simulation experiments show that the improved genetic operators to improve the convergence rate, to be able to search the global optimal solution; on the fitness function designed to properly distinguish the individual strengths and weaknesses so that the outstanding individuals will not spread too fast, poor individual will not disappear soon, will help maintain the diversity of groups was an effective way to prevent premature groups.According to the analysis of the demand,established the Harbin Engineering University undergraduate bank management system. The module of proposed paper is the core of the system, according to the improved method of intelligence composed paper can be directly generated Word form of papers and the papers' answer , papers and answer can be edited, modified, printed in the Word .The results show that the new method to improve the success rate in the test paper and the speed of convergence have been markedly improved, better to overcome the phenomenon of premature convergence.
Keywords/Search Tags:GA, Bound goals, Combinatorial optimization, Smart Paper, Mathematical model
PDF Full Text Request
Related items