消除局域分解端部效应的BP神经网络闭合方法

发布时间：2018-02-23 21:26

  本文关键词： 局部均值分解 BP神经网络 仿真信号 端部效应　出处：《电子技术应用》2017年05期 　论文类型：期刊论文

【摘要】：详细阐述了局部均值分解(LMD)信号处理方法,该方法非常适合处理非平稳信号,可其端部效应严重制约了其进一步应用推广。镜像延拓是局域分解端部效应处理的基本途径,需要镜像面放置在局部极值点处,而实际信号有时难以满足这个条件,可能导致信号分解结果严重失真现象。为此,提出了一种基于传统镜像延拓与BP神经网络相结合进行信号预测以改进LMD端部效应消除效果的新方法。通过BP神经网络模型预测原始信号端点之外的数据点,由此捕捉到端点之外的一个或者多个极值点,再用镜像技术形成闭合处理,从而抑制端部效应。仿真信号的应用实例表明,所提方法可以有效抑制LMD端部效应。
[Abstract]:The method of local mean decomposition (LMD) signal processing is described in detail. This method is very suitable for dealing with non-stationary signals, but its end effect seriously restricts its further application and popularization. Image continuation is the basic way to deal with local decomposition end effects. The image plane needs to be placed at the local extremum, and the actual signal is sometimes difficult to satisfy this condition, which may lead to serious distortion of the signal decomposition result. A new method of signal prediction based on the combination of traditional image continuation and BP neural network is proposed to improve the effect of eliminating the end effect of LMD. The BP neural network model is used to predict the data points outside the original signal endpoint. One or more extreme points outside the endpoint are captured, and then closed processing is formed by mirror image technique to suppress the end effect. The application of the simulation signal shows that the proposed method can effectively suppress the end effect of LMD.
【作者单位】： 江苏第二师范学院数学与信息技术学院;江苏第二师范学院信息化建设与管理办公室;南京南瑞集团信息通信技术分公司;
【基金】：国家自然科学基金(61272506) 国家科技支撑计划课题(2007BAB18B01)
【分类号】：TN911.7;TP183
，

本文编号：1527686