Hi all, I have signals that are very chaotic and choppy. I suspect that there are quite a bunch of major and minor cycles inside. What's the best way to detect and single out those cycles? Any Matlab code around? Thanks a lot!
what's the best way to detect cycle in time series?
Started by ●June 11, 2010
Reply by ●June 11, 20102010-06-11
On 6/11/2010 2:13 PM, Luna Moon wrote:> Hi all, > > I have signals that are very chaotic and choppy. I suspect that there > are quite a bunch of major and minor cycles inside. What's the best > way to detect and single out those cycles?FFT?> Any Matlab code around?Sure! Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●June 11, 20102010-06-11
On Jun 11, 2:24�pm, Jerry Avins <j...@ieee.org> wrote:> On 6/11/2010 2:13 PM, Luna Moon wrote: > > > Hi all, > > > I have signals that are very chaotic and choppy. I suspect that there > > are quite a bunch of major and minor cycles inside. What's the best > > way to detect and single out those cycles? > > FFT? > > > Any Matlab code around? > > Sure! > > Jerry > -- > Engineering is the art of making what you want from things you can get.How? I am looking for a complete solution...
Reply by ●June 11, 20102010-06-11
Luna Moon wrote:> Hi all, > > I have signals that are very chaotic and choppy. I suspect that there > are quite a bunch of major and minor cycles inside. What's the best > way to detect and single out those cycles? > > Any Matlab code around? > > Thanks a lot!Wow, what a cretin!
Reply by ●June 11, 20102010-06-11
On 06/11/2010 12:05 PM, Luna Moon wrote:> On Jun 11, 2:24 pm, Jerry Avins<j...@ieee.org> wrote: >> On 6/11/2010 2:13 PM, Luna Moon wrote: >> >>> Hi all, >> >>> I have signals that are very chaotic and choppy. I suspect that there >>> are quite a bunch of major and minor cycles inside. What's the best >>> way to detect and single out those cycles? >> >> FFT? >> >>> Any Matlab code around? >> >> Sure! >> >> Jerry >> -- >> Engineering is the art of making what you want from things you can get. > > How? I am looking for a complete solution...There's a reason that 'work' is a dirty four letter word. FFT, look for significant spikes. What's "significant" in your case will be different from the next guy's -- which is why you have to do the work. -- Tim Wescott Control system and signal processing consulting www.wescottdesign.com
Reply by ●June 11, 20102010-06-11
On Jun 12, 7:27�am, Tim Wescott <t...@seemywebsite.now> wrote:> On 06/11/2010 12:05 PM, Luna Moon wrote: > > > > > On Jun 11, 2:24 pm, Jerry Avins<j...@ieee.org> �wrote: > >> On 6/11/2010 2:13 PM, Luna Moon wrote: > > >>> Hi all, > > >>> I have signals that are very chaotic and choppy. I suspect that there > >>> are quite a bunch of major and minor cycles inside. What's the best > >>> way to detect and single out those cycles? > > >> FFT? > > >>> Any Matlab code around? > > >> Sure! > > >> Jerry > >> -- > >> Engineering is the art of making what you want from things you can get. > > > How? I am looking for a complete solution... > > There's a reason that 'work' is a dirty four letter word. > > FFT, look for significant spikes. �What's "significant" in your case > will be different from the next guy's -- which is why you have to do the > work. > > -- > Tim Wescott > Control system and signal processing consultingwww.wescottdesign.comIn ocean wave analysis, we use two approaches: 1. Spectral analysis - "significant wave height" Hs is 4*sqrt(M0), where M0 is the zeroth moment of the spectrum (i.e. the variance) and the period is usually taken as the peak of the spectrum (but there are others). 2. Zero-crossing analysis - split the signal into a series of waves between zero up-crossings, then sort the waves and calculate the average of the highest third of the waves, this is Hz. Tz is the average period of the highest third of the waves. Hz should be close to Hs or there is something screwy with your signal.
Reply by ●June 11, 20102010-06-11
On 06/11/2010 02:18 PM, TideMan wrote:> On Jun 12, 7:27 am, Tim Wescott<t...@seemywebsite.now> wrote: >> On 06/11/2010 12:05 PM, Luna Moon wrote: >> >> >> >>> On Jun 11, 2:24 pm, Jerry Avins<j...@ieee.org> wrote: >>>> On 6/11/2010 2:13 PM, Luna Moon wrote: >> >>>>> Hi all, >> >>>>> I have signals that are very chaotic and choppy. I suspect that there >>>>> are quite a bunch of major and minor cycles inside. What's the best >>>>> way to detect and single out those cycles? >> >>>> FFT? >> >>>>> Any Matlab code around? >> >>>> Sure! >> >>>> Jerry >>>> -- >>>> Engineering is the art of making what you want from things you can get. >> >>> How? I am looking for a complete solution... >> >> There's a reason that 'work' is a dirty four letter word. >> >> FFT, look for significant spikes. What's "significant" in your case >> will be different from the next guy's -- which is why you have to do the >> work. >> >> -- >> Tim Wescott >> Control system and signal processing consultingwww.wescottdesign.com > > In ocean wave analysis, we use two approaches: > 1. Spectral analysis - "significant wave height" Hs is 4*sqrt(M0), > where M0 is the zeroth moment of the spectrum (i.e. the variance) and > the period is usually taken as the peak of the spectrum (but there are > others). > 2. Zero-crossing analysis - split the signal into a series of waves > between zero up-crossings, then sort the waves and calculate the > average of the highest third of the waves, this is Hz. Tz is the > average period of the highest third of the waves. > > Hz should be close to Hs or there is something screwy with your > signal.Which is an example of a way to do it in a specific domain, that has a specific probability distribution to the signal. Other signals (a few simple tones of equal amplitude buried in wideband noise, for instance) will have other methods that would need to be applied. Which is why we're all telling the OP that there isn't a prepackaged solution, unless it's specific to his problem. -- Tim Wescott Control system and signal processing consulting www.wescottdesign.com
Reply by ●June 15, 20102010-06-15
Dnia 11-06-2010 o 21:10:35 Vladimir Vassilevsky <nospam@nowhere.com> napisa�(a):> > > Luna Moon wrote: > >> Hi all, >> I have signals that are very chaotic and choppy. I suspect that there >> are quite a bunch of major and minor cycles inside. What's the best >> way to detect and single out those cycles? >> Any Matlab code around? >> Thanks a lot! > > Wow, what a cretin!Yes, loonatic moonthinking humanoid thing. You were very precise, thank you. Is there any cretin filtering tool? -- Mikolaj
Reply by ●June 16, 20102010-06-16
On Jun 11, 2:13�pm, Luna Moon <lunamoonm...@gmail.com> wrote:> Hi all, > > I have signals that are very chaotic and choppy. I suspect that there > are quite a bunch of major and minor cycles inside. What's the best > way to detect and single out those cycles? > > Any Matlab code around? > > Thanks a lot!Forget about FFT (and the rest of regular folks in this group, especially VLV) Recurrence analysis is usually the best for chaotic signals http://en.wikipedia.org/wiki/Recurrence_plot I believe you can find plenty of matlab code if you google
Reply by ●June 16, 20102010-06-16
Mikolaj wrote:> Dnia 11-06-2010 o 21:10:35 Vladimir Vassilevsky <nospam@nowhere.com> > napisa�(a): > >> >> >> Luna Moon wrote: >> >>> Hi all, >>> I have signals that are very chaotic and choppy. I suspect that there >>> are quite a bunch of major and minor cycles inside. What's the best >>> way to detect and single out those cycles? >>> Any Matlab code around? >>> Thanks a lot! >> >> >> Wow, what a cretin! > > > Yes, loonatic moonthinking humanoid thing. > You were very precise, thank you. > Is there any cretin filtering tool?Cretin = gmail & (no real name) & matlab VLV
