期刊文献+
共找到2篇文章
< 1 >
每页显示 20 50 100
非球形分布数据集的去噪方法
1
作者 张岩 闫德勤 郑宏亮 《计算机应用》 CSCD 北大核心 2011年第10期2786-2789,共4页
针对传统支持向量机(SVM)对噪声点过于敏感,模糊支持向量机(FSVM)又对样本集几何形状过分依赖等问题,提出基于噪声过滤系统的粗糙支持向量机(NFS-RSVM)。该方法首先用噪声过滤系统(NFS)将极可能为噪声点的样本过滤掉;然后将数据间隐含... 针对传统支持向量机(SVM)对噪声点过于敏感,模糊支持向量机(FSVM)又对样本集几何形状过分依赖等问题,提出基于噪声过滤系统的粗糙支持向量机(NFS-RSVM)。该方法首先用噪声过滤系统(NFS)将极可能为噪声点的样本过滤掉;然后将数据间隐含的等价类信息作为双惩戒因子融入到支持向量机模型中,进一步区分有效样本和噪声样本。基于UCI数据集的仿真结果表明,NFS-RSVM方法能有效地将数据中的大部分噪声点去除,与传统的SVM和FSVM相比分类精度有一定程度的提高。因此,该方法在处理噪声样本较多又呈现非球形分布的数据集时,表现出较好的抗噪性、分类效果和泛化能力。 展开更多
关键词 支持向量机 粗糙支持向量机 噪声过滤系统 等价类 去噪
下载PDF
Hilbert-Huang transform based noise filtering for the identification of structural systems
2
作者 LIN Jeng-wen HUANG Chih-wei 《Journal of Civil Engineering and Architecture》 2009年第3期1-10,共10页
This paper proposes an enhanced noise filtering technique to the Hilbert-Huang transformation technology for the identification of structural systems using vibration measurements. The Hilbert-Huang transformation tech... This paper proposes an enhanced noise filtering technique to the Hilbert-Huang transformation technology for the identification of structural systems using vibration measurements. The Hilbert-Huang transformation technology is a set of superior algorithms non-stationary signals and also for analyzing non-linear and offers increased accuracy for linear and stationary signals. However, the signals are filtered by reconstruction from "'selected" intrinsic mode functions (IMFs), derived from the original signal through the empirical mode decomposition method. The proposed filtering technique offers the criterion for selecting the IMFs using the orthogonalization coefficient. In addition, a simple free vibration modal analysis has been resolved for the evaluation of the modal damping ratio. Through the enhanced filtering, it is possible to increase the accuracy for the estimation of the time-varying system's natural frequency and the damping ratio, indicative of the degree of damage, which helps an effective design of a control system. 展开更多
关键词 Hilbert-Huang transformation empirical mode decomposition orthogonalization coefficient modal analysis
下载PDF
上一页 1 下一页 到第
使用帮助 返回顶部