Spectrum analysis by autoregressive methods: Performance on application to stationary signals
In order to develop a method capable of determining the time variant spectrum of time series, various existing approaches have been investigated. Although the Fourier-based methods are superior in their computational efficiency, their inherent characteristics may sometimes limit applications. The AR...
Saved in:
Main Authors: | , |
---|---|
其他作者: | |
格式: | Article |
語言: | English |
出版: |
The Japan Society of Mechanical Engineers
2011
|
主題: | |
在線閱讀: | http://kb.psu.ac.th/psukb/handle/2010/7191 |
標簽: |
添加標簽
沒有標簽, 成為第一個標記此記錄!
|
機構: | Prince of Songkhla University |
語言: | English |
總結: | In order to develop a method capable of determining the time variant spectrum of time series, various existing approaches have been investigated. Although the Fourier-based methods are superior in their computational efficiency, their inherent characteristics may sometimes limit applications. The AR method gives the best results even for small data sets. However, insufficient information is available for determining its applicability. In this report, a brief review, as well as the performance, of various AR methods applied to a certain class of stationary time series is systematically documented. The covariance method is found to be the best solution for the determination of AR coefficients, and many trials using sinusoidal data sets indicate the usefullness and applicability of AR-based spectrum analysis. |
---|