摘要
针对传统的灰色预测模型预测精度随步长的增加而逐渐降低的问题,该文采用一种新的阈值及阈值函数对原始数据(信号)进行小波去噪处理,以期增强建模数据(信号)的可靠性,延长模型的预测步长:通过与传统的硬阈值(函数)和软阈值(函数)的比较,验证新阈值及函数对信号的去噪效果;用去噪处理前后的数据分别建立灰预测模型,通过相应的检验指标来判断两函数的预测精度高低。实验结果表明,该阈值及函数与灰预测的组合预测方法是完全可行的,能够在一定程度上提高模型预测的精度和步长。
Aiming at the problem that the predicting accuracy of traditional grey prediction model grad- ually decreases with the increase of step length, the paper presented a wavelet de-noising method using a new thresholding function to manipulate the original signal, in order to enhance the reliability of signal and extend the predictive steps： by comparing with hard thresholding function and Soft thresholding func- tion, the de-noising effect of the new thresholding function was verified; the grey prediciton models of both before and after de-noising data were established, and the predition accuracy of the two functions was compared by corresponding test index. Experimental result showed that the combined prediction method could help improve the model prediction accuracy and step length.
出处
《测绘科学》
CSCD
北大核心
2016年第3期145-149,共5页
Science of Surveying and Mapping
关键词
小波变换
阈值
灰预测
组合
MATLAB
wavelet transform
thresholding
grey prediction
combined
MATLAB
