摘要
针对在保证一定建模精度的前提下如何提高建模速度这一问题,基于小波分析和核方法理论,提出了一种小波融合核的建模方法。该方法首先利用小波对样本数据序列进行多尺度分解,然后对重构后的近似序列和细节序列分别利用核方法进行回归,最后进行结果的融合。该方法具有小波多分辨率分析和核方法对输入维数不敏感的特点,理论上在保证建模精度的前提下,有更快的建模速度。在此基础上,分别通过一维函数和化工生产数据进行了仿真研究,仿真结果验证了算法的有效性。
Aiming at the problem of improving modeling speed, on the premise of ensuring satisfied modeling preci- sion,based on the theories of wavelet analysis and kernel method, a wavelet fusion kernels modeling method was proposed. It performed multiple-scaled decomposition on sample data series using wavelet transform first/y, then the reconstructed approximate series and detail series were regressed depending on kernel method, the outputs were fused finally. The method owning the traits of wavelet-multiresolution analysis and kernel method＇s insensitiveness to- ward the input dimension has faster modeling speed on the premise of ensuring satisfied modeling precision. On this basis, the simulations were carried out through one dimensional function and the data from the chemical process. The simulation results also show the effectiveness of the proposed algorithm.
出处
《自动化与仪表》
北大核心
2009年第6期30-34,共5页
Automation & Instrumentation
关键词
建模
小波分析
核方法
多尺度
再生核
modeling
wavelet analysis
kernel method
multiple-scaled
reproducing kernel
