| DELTA势阱改进QPSO优化BP算法及其应用 |
| |
| 作者姓名： | 于凤玲 周扬 陈建宏 周汉陵 |
| |
| 作者单位： | 1. 五邑大学经济管理学院, 广东 江门 529020;
2. 湖南文理学院资源环境与旅游学院, 湖南 常德 415000;
3. 中南大学资源与安全工程学院, 长沙 410083 |
| |
| 基金项目： | 国家自然科学基金(51374242)资助 |
| |
| 摘 要： | 为了改进BP算法预测性能,提出QPSO-BP模型.该模型采用DELTA势阱改进的量子粒子群(QPSO)算法优化BP网络的权值与阈值,然后利用各年的GDP数据进行训练和预测.结果表明:经过DELTA势阱改进的QPSO优化BP算法模型比PSO-BP模型和BP神经网络更稳定,预测精度更高且泛化能力更强.与文献中所用模型的运算结果相比较,这种改进模型运算结果的相对误差和平均误差更小,在准确性上也有一定的优势.
|
| 收稿时间： | 2013-10-16 |
| 修稿时间： | 2014-07-28 |
BP neural network optimized with QPSO algorithm improved by DELTA potential trough and its application |
| |
| Authors: | YU Fengling ZHOU Yang CHEN Jianhong ZHOU Hanling |
| |
| Institution: | 1. School of Economics & Management, Wu Yi University, Jiangmen 529020, Guangdong, China;
2. College of Resources and Environment and Tourism, Hunan University of Arts and Science, Changde 415000, Hunan, China;
3. School of Resources and Safety Engineering, Central South University, Changsha 410083, China |
| |
| Abstract: | To improve the generalization ability of BP network for prediction, a BP neural network optimized with QPSO is proposed. This model uses the QPSO improved by δ potential trough to optimize the initial values of weights and thresholds of BP network. Then the data of each year's GDP are selected in training and prediction. The experiments show that the QPSO-BP network optimized by using δ potential trough produces stable prediction results. Compared with the prediction models of PSO-BP and BP, the proposed model has a better generalization ability and a higher accuracy. In addition, the calculation results of the improved QPSO-BP optimization algorithm model have smaller relative errors and average errors compared with the results of the models in the literature. |
| |
| Keywords: | back-propagation neural network PSO model QPSO model δ potential trough" target="_blank">δ potential trough')">δ potential trough GDP |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《》浏览原始摘要信息 |
| 点击此处可从《》下载免费的PDF全文 |
|
