集成GASA混合学习策略的BP神经网
络在水稻虫害预测中的应用 |
| |
作者姓名: | 汪 璇 吕家恪 谢德体 魏朝富 |
| |
作者单位: | 西南大学计算机与信息科学学院,重庆市数字农业重点实验室 重庆400716,西南大学计算机与信息科学学院,重庆市数字农业重点实验室,重庆400716,重庆市数字农业重点实验室,重庆市数字农业重点实验室,重庆400716,重庆400716,重庆400716,重庆400716 |
| |
基金项目: | 西南大学校科研和教改项目 |
| |
摘 要: | 为改进受多变量、时变和不确定因素影响的作物虫情预测的效率和准确性,以重庆市永川区为研究区域,将1996-2004年的气象和虫害数据分为独立的训练和检验数据集,以研究集成GASA混合学习策略的BP神经网络应用于水稻虫情预测中的可行性。与广泛应用的时间序列分析法和传统BP神经网络相比,集成混合学习策略的BP神经网络预测精度和拟合性能都有了很大提高,因而具有较好的应用前景。
|
关 键 词: | 人工神经网络 遗传算法 模拟退火 水稻虫害预测 |
Application of BP Networks Based on GASA Hybrid Strategy to Rices Pests Prediction |
| |
Authors: | Wang Xuan Lv JiaKe Xie DeTi Wei Chaofu |
| |
Institution: | 1.College of Computer and Information Science, Southwest University, Chongqing 400716; 2.Key Laboratory of Digital agriculture, Chongqing 400716 |
| |
Abstract: | To improve pests'' prediction efficiency and precision, Laifu town of Yongchuan, Chongqing was selected for the study. Local meteorological and pets data from 1996-2004 were divided into training and validation datum sets, Rice prediction applied by GASA-based BPNN was operated. Compared with general used time series and BP Neuron Network prediction, the accuracy achieved by GASA-based BPNN was much better. The potential ability of GASA-based BPNN was also discussed. |
| |
Keywords: | Artificial neuron network Genetic Algorithm Simulated annealing algorithm Rice pets occurrence level prediction |
|
| 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 |