Wavelet: Linking Scale coefficients -> TimeStamp in DWT decomposition of time series

29 Views Asked by At

I'm using PyWavelets for singularity detection in a time series of measured values.

The values were taken with a sample time of 10 ms and the length of each data set is 2048 values.

Along with the data set, I have an array with the time stamps for each measured value.

I did

wavelet_name = "bior5.5"
w = pywt.Wavelet(wavelet_name)
maxlev = pywt.dwt_max_level(len(pressure_values), w.dec_len)
coeffs = pywt.wavedec(pressure_values, wavelet_name, level=maxlev)

That leads to coeffs with the following lengths per scale:

scale  1     length =  26
scale  2     length =  42
scale  3     length =  74
scale  4     length =  138
scale  5     length =  265 
scale  6     length =  520
scale  7     length =  1029

I can now find the singularities in the coefficients that appear over multiple scales, but I need to properly link these to a time stamp.

How can this be done properly?

I'm also confused, because I expected that the length of e.g. scale 7 would be 1024 after the downsampling in the decomposition process. With that length it would have been clear which coefficient maps to which time stamp.

Any hint is appreciated!

0

There are 0 best solutions below