research/mysignal
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

upd

Browse source
This commit is contained in:
François Boulogne 2025-09-11 18:10:10 +02:00
parent 19cf916db9
commit cdc5d7a9b9
2 changed files with 30 additions and 26 deletions
Download patch file Download diff file Expand all files Collapse all files

2
mysignal/__init__.py
View file

@ -1 +1 @@
__version__ = '0.1.18'
__version__ = '0.1.19'

54
mysignal/phasefreq.py
View file

@ -70,7 +70,7 @@ def filter_signal_by_modes(time, signal, num_modes=1, bandwidth_factor=0.1, requ
bandwidth = bandwidth_factor * freq
lowcut = max(0.01, freq - bandwidth) # Éviter les fréquences négatives
highcut = freq + bandwidth
# Éviter les fréquences supérieures à la fréquence de Nyquist
nyq = 0.5 * fs
highcut = min(highcut, nyq * 0.99) # Laisser une marge
@ -78,18 +78,18 @@ def filter_signal_by_modes(time, signal, num_modes=1, bandwidth_factor=0.1, requ
# Normaliser les fréquences
lowcut_norm = lowcut / nyq
highcut_norm = highcut / nyq
print(f"Filtrage à {freq:.2f} Hz, bande : [{lowcut_norm:.5f}, {highcut_norm:.5f}] (normalisé)")
# Vérifier si le signal est suffisamment long
period = 1 / lowcut if lowcut > 0 else 0
#required_cycles = 40 # Vous avez mentionné 40 cycles
min_length = required_cycles * period * fs
if len(signal_clean) < min_length:
print(f"Avertissement: Signal court pour {freq:.2f} Hz. "
f"Requis: {min_length:.0f} échantillons, Disponible: {len(signal_clean)}")
# Méthode alternative: utiliser un filtre SOS (Second-Order Sections)
# plus stable pour les signaux courts
sos = butter(4, [lowcut_norm, highcut_norm], btype='band', output='sos')
@ -98,7 +98,7 @@ def filter_signal_by_modes(time, signal, num_modes=1, bandwidth_factor=0.1, requ
# Méthode standard
b, a = butter(4, [lowcut_norm, highcut_norm], btype='band')
filtered_signal = filtfilt(b, a, signal_clean)
filtered_signals.append(filtered_signal)
# Le temps reste le même après filtrage
@ -107,17 +107,17 @@ def filter_signal_by_modes(time, signal, num_modes=1, bandwidth_factor=0.1, requ
return np.array(filtered_signals), dominant_freqs, time_filtered
def plot_filtered_modes(t, e, e_filtered, e_frequencies, e_time, n_modes):
def plot_filtered_modes(t, e, e_filtered, e_frequencies, e_time, n_modes, output=None):
fig, ax = plt.subplots(nrows=n_modes+1, ncols=2, figsize=(12, 4 * n_modes), gridspec_kw={'width_ratios': [3, 1]})
ax[0, 0].plot(t, e)
ax[0, 0].set_title('Signal original')
ax[0, 0].set_xlabel('Temps (s)')
ax[0, 0].set_ylabel('Amplitude')
for i in range(n_modes):
ax[i+1, 0].plot(e_time, e_filtered[i])
ax[i+1, 0].set_title(f'Mode #{i+1} filtré ({e_frequencies[i]:.2f} Hz)')
@ -126,10 +126,12 @@ def plot_filtered_modes(t, e, e_filtered, e_frequencies, e_time, n_modes):
ax[i+1, 1].plot(e_time, e, color='red')
ax[i+1, 1].plot(e_time, e_filtered[i])
ax[i+1, 1].set_xlim(left=0.5 * e_time.mean(), right=0.5*e_time.mean() + 10/e_frequencies[i])
plt.tight_layout()
plt.show()
if output:
plt.savefig(output)
@ -238,7 +240,7 @@ def analyze_signal_Hilbert(time, e_signal, s_signal, freq_rtol=0.01):
period_e = 1 / freq_e
period_s = 1 / freq_s
res = {'period_e': period_e,
'period_s': period_s,
@ -247,7 +249,7 @@ def analyze_signal_Hilbert(time, e_signal, s_signal, freq_rtol=0.01):
'phase': mean_phase_diff,
'phrase_deg': phase_diff_deg,
'delay': time_shift}
return res
def analyze_signal_sinfit(time, e_signal, s_signal, freq_rtol=0.01):
@ -344,7 +346,7 @@ def analyze_signal_sinfit(time, e_signal, s_signal, freq_rtol=0.01):
period_e = 1 / freq_e
period_s = 1 / freq_s
res = {'period_e': period_e,
'period_s': period_s,
'freq_e': freq_e,
@ -352,7 +354,7 @@ def analyze_signal_sinfit(time, e_signal, s_signal, freq_rtol=0.01):
'phase': phase_diff,
'phrase_deg': phase_diff_deg,
'delay': time_shift}
return res
@ -460,7 +462,7 @@ def analyze_signal_cross_correlation(time, e_signal, s_signal, freq_rtol=0.01):
'phase': phase_diff,
'phrase_deg': phase_diff_deg,
'delay': time_shift}
return res
@ -571,18 +573,18 @@ def analyze_signal_wavelet(time, e_signal, s_signal, freq_rtol=0.01):
'phase': mean_phase_diff,
'phrase_deg': phase_diff_deg,
'delay': time_shift}
return res
def plot_phases(e_time, e_filtered, e_frequencies, s_filtered, n_modes, callback=analyze_signal_wavelet):
fig, ax = plt.subplots(nrows=n_modes, ncols=2, figsize=(12, 6), gridspec_kw={'width_ratios': [3, 1]})
for mod in range(n_modes):
res = callback(e_time, e_filtered[mod], s_filtered[mod], freq_rtol=0.3)
ax[mod, 0].set_title(f'Freq: {res['freq_e']:.3f}, Phase: {res['phase']:.3f}, Delay: {res['delay']:.3f}' )
ax[mod, 0].plot(e_time, e_filtered[mod], label='e')
ax[mod, 0].plot(e_time, s_filtered[mod], label='s')
@ -592,9 +594,11 @@ def plot_phases(e_time, e_filtered, e_frequencies, s_filtered, n_modes, callback
ax[mod, 1].plot(e_time, e_filtered[mod], label='e')
ax[mod, 1].plot(e_time, s_filtered[mod], label='s')
ax[mod, 1].set_xlim(left=0.5 * e_time.mean(), right=0.5*e_time.mean() + 10/e_frequencies[mod])
for a in ax[:, 0]:
a.legend()
plt.tight_layout();
plt.show();
plt.show();
if output:
plt.savefig(output)