基于改进遗传算法的智能测验系统研究.rar

摘要:近几年智能优化算法倍受人们关注，并在诸多领域取得了成功。组卷问题是一个在一定约束条件下的多目标参数优化问题，传统的组卷算法具有组卷速度慢、成功率较低、试卷质量不高等缺点。随着计算机辅助教学和人工智能的发展,大型题库系统中，能决定组卷的质量和效率的组卷算法逐渐被众多专家所关注。

　　　本论文根据应用型本科院校《自动控制理论》课程特点，建立了试题库，并通过对学生调查研究对试题属性进行了合理赋值。采用权重系数法将多目标优化问题转化为单目标优化问题，建立了目标函数和数学模型。针对单纯遗传算法存在许多缺点如早熟收敛、局部搜索能力不强等，本文编程实现了一种基于粒子群算法的改进遗传算法并将其应用于智能组卷中，通过设置不同的参数在MATLAB平台上进行仿真实验。仿真实验的结果表明，改进遗传算法应用于《自动控制理论》智能组卷是合理可行的，可以有效节省教学资源，提高工作效率。

关键词 智能优化；自动控制理论；数学模型；遗传算法；早熟收敛

Abstract:In recent years, much attention is attached on intelligent optimization algorithms. They have been used successfully in many fields. Testing paper is a certain constraints under the multi-objective parameter optimization problem, the shortcomings of traditional algorithmic of testing paper are so obvious. The rate is very low and the quality of the paper is so poor. With the development of computer-assisted instruction and artificial intelligence, the algorithms that can determine the quality and efficiency of testing paper gradually attract the attention of many experts in large bank system.

　　　 The test database focusing on the characteristics of Automatic Control Theory course was established, and questions were reasonably assigned based on the research to the students in my thesis. The multi-objective optimization problems are turned into a single objective optimization problem by using weight coefficient method, the objective function and mathematical model are established meanwhile. A new algorithm that can solve the shortcomings of traditional genetic algorithm based on particle swarm optimization algorithm was used in this paper. For example, premature convergence and the local search ability is poor. The improved algorithm is applied in intelligent test, and it is validated by the simulation of MATLAB with changing the parameters. Simulation results show that, the new algorithm is feasible when used in testing paper. It can save the teaching resources and improve the work efficiency.

Keywords  Intelligent Optimization  Automatic Control Theory  Mathematical Model Genetic Algorithm  Premature Convergence