def __init__(self,
signal_in: np.ndarray,
wavelet_in: str = 'dog',
order: int = 2,
padding: str = 'none',
frequency_resolution: float = 0.5) -> None:
"""
Parameters
----------
signal_in : numpy.ndarray
1D input signal.
wavelet_in : str
Wavelet function ('dog' or 'morlet').
order : int
Order of the wavelet function.
padding : str
Padding method ('zero', 'mirror', or 'none').
frequency_resolution : float
Wavelet space resolution in scale/frequency.
Returns
-------
NoneType
None
"""
# save input data
self.m_supported_wavelets = ['dog', 'morlet']
# check supported wavelets
if wavelet_in not in self.m_supported_wavelets:
raise ValueError(f'Wavelet {wavelet_in} is not supported')
if wavelet_in == 'dog':
self._m_c_reconstructions = {2: 3.5987,
4: 2.4014,
6: 1.9212,
8: 1.6467,
12: 1.3307,
16: 1.1464,
20: 1.0222,
30: 0.8312,
40: 0.7183,
60: 0.5853}
elif wavelet_in == 'morlet':
self._m_c_reconstructions = {5: 0.9484,
6: 0.7784,
7: 0.6616,
8: 0.5758,
10: 0.4579,
12: 0.3804,
14: 0.3254,
16: 0.2844,
20: 0.2272}
self.m_wavelet = wavelet_in
if padding not in ['none', 'zero', 'mirror']:
raise ValueError('Padding can only be none, zero or mirror')
self._m_data = signal_in - np.ones(len(signal_in)) * np.mean(signal_in)
self.m_padding = padding
self.__pad_signal()
self._m_data_size = len(self._m_data)
self._m_data_mean = np.mean(signal_in)
if order not in self._m_c_reconstructions:
raise ValueError('Wavelet ' + str(wavelet_in) + ' does not support order '
+ str(order) + ". \n Only orders: " +
str(sorted(self._m_c_reconstructions.keys())).strip('[]') +
" are supported")
self.m_order = order
self._m_c_final_reconstruction = self._m_c_reconstructions[order]
# create scales for wavelet transform
self._m_scales = autoscales(N=self._m_data_size,
dt=1,
dj=frequency_resolution,
wf=wavelet_in,
p=order)
self._m_number_of_scales = len(self._m_scales)
self._m_frequency_resolution = frequency_resolution
self.m_spectrum = None
def __init__(self,
signal_in,
wavelet_in='dog',
order=2,
padding="none",
frequency_resolution=0.5):
"""
Constructor of the WaveletAnalysisCapsule
:param signal_in: A 1d input signal
:param wavelet_in: mother wavelet function can be either "dog" or "morlet"
:type wavelet_in: str
:param order: order of the mother wavelet function
:param padding: padding strategy. Zero and mirror padding ara supported.
:param frequency_resolution: wavelet space resolution in scale / frequency
"""
# save input data
self.m_supported_wavelets = ['dog', 'morlet']
# check supported wavelets
if wavelet_in not in self.m_supported_wavelets:
raise ValueError('Wavelet ' + str(wavelet_in) +
' is not supported')
if wavelet_in == 'dog':
self._m_c_reconstructions = {
2: 3.5987,
4: 2.4014,
6: 1.9212,
8: 1.6467,
12: 1.3307,
16: 1.1464,
20: 1.0222,
30: 0.8312,
40: 0.7183,
60: 0.5853
}
elif wavelet_in == 'morlet':
self._m_c_reconstructions = {
5: 0.9484,
6: 0.7784,
7: 0.6616,
8: 0.5758,
10: 0.4579,
12: 0.3804,
14: 0.3254,
16: 0.2844,
20: 0.2272
}
self.m_wavelet = wavelet_in
if padding not in ["none", "zero", "mirror"]:
raise ValueError("Padding can only be none, zero or mirror")
self._m_data = signal_in - np.ones(len(signal_in)) * np.mean(signal_in)
self.m_padding = padding
self.__pad_signal()
self._m_data_size = len(self._m_data)
self._m_data_mean = np.mean(signal_in)
if order not in self._m_c_reconstructions:
raise ValueError(
'Wavelet ' + str(wavelet_in) + ' does not support order ' +
str(order) + ". \n Only orders: " +
str(sorted(self._m_c_reconstructions.keys())).strip('[]') +
" are supported")
self.m_order = order
self._m_c_final_reconstruction = self._m_c_reconstructions[order]
# create scales for wavelet transform
self._m_scales = autoscales(N=self._m_data_size,
dt=1,
dj=frequency_resolution,
wf=wavelet_in,
p=order)
self._m_number_of_scales = len(self._m_scales)
self._m_frequency_resolution = frequency_resolution
self.m_spectrum = None
return