基于人工蜂群改进的BP神经网络移动用户行为分析及预测方法

发布时间：2018-03-01 01:32

本文关键词： BP神经网络人工蜂群移动用户行为分析预测 Matlab　出处：《沈阳农业大学学报》2015年06期 　论文类型：期刊论文

【摘要】：如何根据不同的用户行为,来为移动用户提供精准的个性化服务是目前移动应用服务开发技术发展的主流。为解决BP神经网络建模算法收敛速度慢及预测不准确问题,提出基于人工蜂群算法改进的BP神经网络算法。为测试改进后算法的准确性,采用Matlab编程进行试验仿真,通过黑盒子测试方法输出预测的用户行为和实际的用户行为。在18次预测中只有2次预测失败,预测成功率达80%以上。为了验证改进的BP神经网络算法的效率,采用初始总群数为1000,进行了收敛性测试。试验结果表明:基于人工蜂群算法改进的BP神经网络算法可以有效的提高移动用户行为分析的效率和准确性,对在使用移动用户行为分析模型构建过程中,准确定位用户上网需求,提升企业在营销中的竞争力具有非常重要的意义。
[Abstract]:In order to solve the problem of slow convergence and inaccurate prediction of BP neural network modeling algorithm, how to provide accurate personalized service for mobile users according to different user behavior is the mainstream of mobile application service development at present. An improved BP neural network algorithm based on artificial bee colony algorithm is proposed. In order to test the accuracy of the improved algorithm, Matlab programming is used to carry out the experimental simulation. In order to verify the efficiency of the improved BP neural network algorithm, the predicted user behavior and actual user behavior were outputted by the black box test method. Only two of the 18 predictions failed, and the prediction success rate was over 80%. The convergence test is carried out with the initial total number of clusters 1000. The experimental results show that the improved BP neural network algorithm based on artificial bee colony algorithm can effectively improve the efficiency and accuracy of mobile user behavior analysis. It is of great significance in the process of building mobile user behavior analysis model to accurately locate users' needs on the Internet and enhance the competitiveness of enterprises in marketing.
【作者单位】：沈阳农业大学信息与电气工程学院;中国移动通信集团辽宁有限公司;
【基金】：国家科技支撑计划项目(2012BAJ26B00) 北京农业信息技术研究中心开放课题项目(2013)
【分类号】：TP18

【参考文献】