摘要
为满足复杂工程预测问题的需要,把粗集理论与正交小波网络相结合,建立了一种基于粗集的正交小波网络预测模型。应用主成分分析方法解决了正交小波网络多维输入时的维数灾难,提高了网络的收敛性和预测的时效性。预测模型兼容了正交小波网络和粗神经网络的优良特性,具有良好的函数逼近能力和极强的鲁棒性,特别适合于具有随机因素的高精度预测问题。仿真研究表明,模型的预测精度和收敛速度优于小波框架神经网络。
Combining the orthogonal wavelet networks with rough sets theory, a forecasting model of the orthogonal wavelet networks based on rough rets is put forward. Using the principal component analysis （PCA） about input vectors, the model keeps away from the dimension avalanche of orthogonal wavelet networks. Combined the excellent characteristics of rough networks and orthogonal wavelet networks together, the model is provided with the favorable robust and function approximating ability. It especially fits to the precise forecasting applications with randomicity. The experiment results show that the model is superior to frame wavelet networks in some aspects of forecasting precision, network convergence and its robust to uncertain factors.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2005年第8期1462-1466,共5页
Systems Engineering and Electronics
基金
山东省中青年科学家发展基金(031BS147)
山东省科技攻关计划(031080112)资助课题
关键词
小波网络
预测
粗集
主成分分析
wavelet network
forecast
rough sets
principal component analysis
