摘要
针对工程应用中的信号缓变特性,提出截断奇异值分解(TSVD)和希尔伯特(Hilbert)变换联合使用的相位差计算方法。首先对离散的含噪一维信号分段并构成汉克尔(Hankel)矩阵,根据TSVD的奇异值依次降序排列的规律,用奇异熵变化量的拐点确定奇异谱阶次,使该阶次内的信号饱和度基本代替原信号。将降噪后的矩阵重构得到降噪后的信号,对降噪后的信号进行Hilbert变换计算两路信号的相位差。仿真分析及实测数据表明:该联合计算方法满足工程应用,且具有一定的通用性。
A method of combining TSVD with Hilbert transformation was introduced,and was used for calculating phase difference of slowly changing signals in engineering application.Firstly,discrete one-dimensional signal was divided into several segments,which are composed into a Hankel matrix.Hankel matrix was decomposed by TSVD,by means of singular entropy variation tendency,finding the optimal inflexion in order to include main signal components.The noise-reduced matrix was then used to reconstruct the noise-reduced signal.According to Hilbert transformation of the noise-reduced signal,phase difference value could be calculated.Simulated analysis and practical data show that this jointed method would be suited for engineering application and also has a certain commonality.
出处
《仪表技术与传感器》
CSCD
北大核心
2014年第6期144-146,149,共4页
Instrument Technique and Sensor
基金
国家自然科学基金(61102115)
广西自然科学基金(2012GXNSFBA053177)
广西研究生教育创新计划资助项目(YCSZ2013067)
关键词
截断奇异值分解
希尔伯特变换
奇异熵
相位差
汉克尔矩阵
truncated singular value decomposition
Hilbert transformation
singular entropy
phase difference
Hankel matrix