| 引用本文： |
-
吴建生.旱涝灾害的遗传-神经网络集成预测方法研究[J].广西科学,2006,13(3):203-206,211. [点击复制]
- WU Jian-sheng.A Study on Genetic Algorithms-Neural Network Ensemble Forecasting Methods of Drought and Water-logging Disasters[J].Guangxi Sciences,2006,13(3):203-206,211. [点击复制]
|
|
| 摘要: |
| 利用遗传算法的全局搜索能力同时进化设计三层BP神经网络的结构和连接权,并以进化后的网络结构和连接权作为新的神经网络结构和初始连接权,再进行新一轮附加动量的BP神经网络训练,把训练后的结果简单平均集成,以此建立旱涝灾害的遗传-神经网络集成预测新方法。应用该方法对广西桂林6月(主汛期1995~2005年)的降水量进行实例预测的结果表明,该方法的收敛速度快,预报精度高,易于操作,是一种具有较高应用价值的预测方法。 |
| 关键词: 旱涝灾害 预测遗传算法 神经网络集成 |
| DOI： |
| 投稿时间：2005-12-19修订日期：2006-03-29 |
| 基金项目:广西科学研究与技术发展计划项目(桂攻关:0592005-2A);广西教育厅项目(200508234) |
|
| A Study on Genetic Algorithms-Neural Network Ensemble Forecasting Methods of Drought and Water-logging Disasters |
|
WU Jian-sheng
|
| (Department of Mathematics and Computer Science, Liuzhou Teacher School, Liuzhou, Guangxi, 545004, China) |
| Abstract: |
| Evolved the neural network architecture and connection weights by using global research ability of genetic algorithms, new neural network architecture and beginning start connection weights be made of the evolution network structure and the connection, and train again the traditional back propagation by training samples and ensemble results by mean, this method be established the forecast new model of drought and water-logging.The application example is build with monthly mean rainfall of Guilin of Guangxi during 1995 to 2005.The calculation result express that our method of forecast can improves convergence speed and forecast accuracy.It is a useful model for forecasting. |
| Key words: drought and water-logging genetic algorithms neural network ensemble |
