摘要
针对传统的遗传算法收敛慢的问题,提出了一种改进的遗传算法并将其应用在学生成绩预测中。所采用的遗传算法改进策略包括:(1)采用实数进行编码;(2)建立个体适应值函数进行个体评价;(3)使用新的选种策略;(4)改进了杂交过程;(5)修改了入选概率小于变异概率的个体变异策略;(6)优化了算法结束条件。本文将BP神经网络和改进的遗传算法相结合构造学生成绩预测模型。实验结果表明,在误差的收敛速度以及成绩预测的准确性方面,本文提出的模型都获得了令人满意的性能。
Considering the problem of slow convergence in traditional genetic algorithm,in this paper,we proposed an improved genetic algorithm and its application in student score predicting.Our improved strategies adopted in genetic algorithms include： 1） Encoding by real number;2） Establishing individual fitness function for individual evaluation;3） Using a novel selection strategy;4） Improving the hybridization process;5） Modifying the individual variation strategy which could select individuals with lower selecting probability than variation probability;6） Optimizing end condition of the proposed algorithm.Afterwards,BP neural network are combined with the improved genetic algorithm to construct student score prediction model.The experimental results show that in the aspects of both error convergence rate and prediction accuracy,the proposed model achieves a satisfactory performance.
出处
《科技通报》
北大核心
2012年第10期223-225,共3页
Bulletin of Science and Technology
基金
2011年新世纪广西壮族自治区高等教育教改工程项目课题(2011JGA273)
关键词
遗传算法
成绩预测
归一化
BP神经网络
genetic algorithm
score prediction
normalization
BP neural network
