def run_i_periodogram(data_sets):
'''
run the periodogram using a cubic spline
'''
sampling_rate = 30
ds_none = list(data_sets.items())[0][1]
ds_none_name = list(data_sets.items())[0][0]
for i in range(1, len(data_sets)):
ds_vibration = list(data_sets.items())[i][1]
ds_vibration_name = list(data_sets.items())[i][0]
fig = plt.figure("Cubic Periodogram " + ds_none_name + " (left), " +ds_vibration_name + " (right)")
gs = fig.add_gridspec(len(ds_none), 2, hspace=1)
axs = gs.subplots()
for i in range(len(ds_none)):
cs = CubicSpline(ds_none[i]["t"], ds_none[i]["ds"])
times = numpy.arange(ds_none[i]["t"][0],ds_none[i]["t"][-1],1000/sampling_rate)
values = cs(times)
p = spectrum.Periodogram(values, sampling=sampling_rate)
p.run()
x, y = p.frequencies(), 10 * spectrum.tools.log10(p.psd)
axs[i][0].plot(x[2:], y[2:])
for i in range(len(ds_none)):
cs = CubicSpline(ds_vibration[i]["t"], ds_vibration[i]["ds"])
times = numpy.arange(ds_vibration[i]["t"][0],ds_vibration[i]["t"][-1],1000/sampling_rate)
values = cs(times)
p = spectrum.Periodogram(values, sampling=sampling_rate)
p.run()
x, y = p.frequencies(), 10 * spectrum.tools.log10(p.psd)
axs[i][1].plot(x[2:], y[2:])
plt.show()
def run_periodogram(data_sets):
'''
Run a periodogram on the data sets
'''
ds_none = list(data_sets.items())[0][1]
ds_none_name = list(data_sets.items())[0][0]
for i in range(1, len(data_sets)):
ds_vibration = list(data_sets.items())[i][1]
ds_vibration_name = list(data_sets.items())[i][0]
fig = plt.figure("Uniform Periodogram " + ds_none_name + " (left), " +ds_vibration_name + " (right)")
gs = fig.add_gridspec(len(ds_none), 2, hspace=1)
axs = gs.subplots()
for i in range(len(ds_none)):
data = ds_none[i]["ds"]
p = spectrum.Periodogram(data, sampling=(len(data)/30))
p.run()
x, y = p.frequencies(), 10 * spectrum.tools.log10(p.psd)
axs[i][0].plot(x[2:], y[2:])
for i in range(len(ds_none)):
data = ds_vibration[i]["ds"]
p = spectrum.Periodogram(data, sampling=(len(data)/30))
p.run()
x, y = p.frequencies(), 10 * spectrum.tools.log10(p.psd)
axs[i][1].plot(x[2:], y[2:])
plt.show()
def updateSpectrum(data):
'''Power spectrum density'''
if not update_data.update_data.rx_stop:
p = spectrum.Periodogram(P.data,
sampling=P.Params['sample_freq'],
window='hann')
p.run()
if sum(p.psd) == 0:
P.PSD = np.zeros(np.size(p.psd))
P.line_spec.set_ydata(P.PSD[P.Params['start']:P.Params['end']])
else:
P.PSD = np.fft.fftshift(10 * np.log10(p.psd))
P.line_spec.set_ydata(P.PSD[P.Params['start']:P.Params['end']])
P.line_spec.set_xdata(P.Params['spec_idx'])
y = np.array(P.PSD[:])
ymax = max(y)
idx = np.where(y > ymax * 0.9)
ave = np.mean(y[idx[0][0]:idx[0][-1]])
threshold = ave / np.sqrt(2)
x = np.where(y > threshold)
bw = (x[0][-1] - x[0][0] +
1) * P.Params['sample_freq'] / 1e3 / P.Params[
'input_sample_len'] #unit: KHz
bandwidth = 'Bandwidth: {:.2f}KHz'.format(bw)
P.bwLabel['text'] = bandwidth
updateCursor()
return P.line_spec,
def timerEvent(self, e): # FFT
global fftbuffersize
if self.datastream == None: return
if self.freeze == 1: return
if SELECTEDCH == BOTH12:
channel = 12
X, Y = self.datastream.read(channel, fftbuffersize, verbose)
if X is None or not len(X): return
data_CH1 = X[:fftbuffersize]
data_CH2 = Y[:fftbuffersize]
elif SELECTEDCH == CH1:
channel = 1
X = self.datastream.read(channel, fftbuffersize, verbose)
if X is None or not len(X): return
data_CH1 = X[:fftbuffersize]
data_CH2 = np.ones((fftbuffersize, ))
elif SELECTEDCH == CH2:
channel = 2
X = self.datastream.read(channel, fftbuffersize, verbose)
if X is None or not len(X): return
data_CH2 = X[:fftbuffersize]
data_CH1 = np.ones((fftbuffersize, ))
self.df = 1.0 / (fftbuffersize * self.dt)
self.setAxisTitle(Qwt.QwtPlot.xBottom,
'Frequency [Hz] - Bin width %g Hz' % (self.df, ))
self.f = np.arange(0.0, samplerate, self.df)
if not SPECTRUM_MODULE:
lenX = fftbuffersize
window = np.blackman(lenX)
sumw = np.sum(window * window)
A = FFT.fft(data_CH1 * window) #lenX
B = (A * np.conjugate(A)).real
A = FFT.fft(data_CH2 * window) #lenX
B2 = (A * np.conjugate(A)).real
sumw *= 2.0 # sym about Nyquist (*4); use rms (/2)
sumw /= self.dt # sample rate
B /= sumw
B2 /= sumw
else:
print "FFT buffer size: %d points" % (fftbuffersize, )
B = spectrum.Periodogram(np.array(data_CH1, dtype=float64),
samplerate)
B.sides = 'onesided'
B.run()
B = B.get_converted_psd('onesided')
B2 = spectrum.Periodogram(np.array(data_CH2, dtype=float64),
samplerate)
B2.sides = 'onesided'
B2.run()
B2 = B2.get_converted_psd('onesided')
if self.logy:
P1 = np.log10(B) * 10.0
P2 = np.log10(B2) * 10.0
P1 -= P1.max()
P2 -= P2.max()
else:
P1 = B
P2 = B2
if not self.average:
self.a1 = P1
self.a2 = P2
self.avcount = 0
else:
self.avcount += 1
if self.avcount == 1:
self.sumP1 = P1
self.sumP2 = P2
elif self.sumP1.shape != P1.shape or self.sumP1.shape != P1.shape:
self.avcount = 1
self.sumP1 = P1
self.sumP2 = P2
else:
self.sumP1 += P1
self.sumP2 += P2
self.a1 = self.sumP1 / self.avcount
self.a2 = self.sumP2 / self.avcount
self.setDisplay()
def create_all_psd():
f = pylab.linspace(0, 1, 4096)
pylab.figure(figsize=(12, 8))
# MA model
p = spectrum.pma(xx, 64, 128)
p()
p.plot()
"""
#ARMA 15 order
a, b, rho = spectrum.arma_estimate(data, 15,15, 30)
psd = spectrum.arma2psd(A=a,B=b, rho=rho)
newpsd = tools.cshift(psd, len(psd)//2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd/max(newpsd)), label='ARMA 15,15')
"""
# YULE WALKER
p = spectrum.pyule(xx, 7, NFFT=4096, scale_by_freq=False)
p.plot()
# equivalent to
# plot([x for x in p.frequencies()] , 10*log10(p.psd)); grid(True)
#burg method
p = spectrum.pburg(xx, 7, scale_by_freq=False)
p.plot()
#pcovar
p = spectrum.pcovar(xx, 7, scale_by_freq=False)
p.plot()
#pmodcovar
p = spectrum.pmodcovar(xx, 7, scale_by_freq=False)
p.plot()
# correlogram
p = spectrum.pcorrelogram(xx, lag=60, NFFT=512, scale_by_freq=False)
p.plot()
# minvar
p = spectrum.pminvar(xx, 7, NFFT=256, scale_by_freq=False)
p.plot()
# pmusic
p = spectrum.pmusic(xx, 10, 4, scale_by_freq=False)
p.plot()
# pmusic
p = spectrum.pev(xx, 10, 4, scale_by_freq=False)
p.plot()
# periodogram
p = spectrum.Periodogram(xx, scale_by_freq=False)
p.plot()
#
legend([
"MA 32", "pyule 7", "pburg 7", "pcovar", "pmodcovar", "correlogram",
"minvar", "pmusic", "pev", "periodgram"
])
pylab.ylim([-80, 80])
N = 1000 # cantidad de muestras df = fs/N # resolución espectral eps = np.finfo(float).tiny # grilla de sampleo temporal tt = np.linspace(0, (N-1)/fs, N).flatten() dd = 0.5 # prestar atención que las tuplas dentro de los diccionarios también pueden direccionarse mediante "ii" #data = np.sin( 2*np.pi*(N/4+dd)*df*tt) + np.sqrt(0.1)*np.random.randn(N) data = np.sin( 2*np.pi*(N/4+dd)*df*tt) + np.sin( 2*np.pi*(N/4+10+dd)*df*tt) + np.sqrt(0.1)*np.random.randn(N) # No paramétricos pP = sp.Periodogram(data, sampling=fs, NFFT = N ) _, pW = sg.welch(data, fs=fs, nfft=N, window='hanning', nperseg=int(np.round(N/3)) ) pBT = sp.pcorrelogram(data, sampling=fs, NFFT = N, lag=int(np.round(N/5)), window='hamming') pMT = sp.MultiTapering(data, sampling=fs, NFFT = N, NW=2 ) # Paramétricos pEig = sp.pev(data, 30, NFFT=N ) pCov = sp.pmodcovar(data, 30, NFFT = N ) pARMA = sp.parma(data, 8, 8, 30, NFFT=N) sp.pma
hamming = Window(len(x), name='hamming')
f, pxx = sci.periodogram(x,
window=hamming.data,
fs=Fs,
nfft=len(x),
scaling='spectrum')
pwrest = pxx.max()
idx = pxx.argmax()
plt.subplot(312)
plt.plot(f, pxx)
plt.title('Periodograma')
plt.xlabel('Frequência')
plt.ylabel('Potência (W)')
plt.grid()
plt.axis([0, 2 * fc, 0, A**2])
print('A potência máxima ocorre em ', f[idx], ' Hz')
print('A potência estimada é', pwrest)
plt.subplot(313)
#construindo todo o procedimento com as funções da spectrum
import spectrum as spec
data = spec.data_cosine(N=len(x), A=10, sampling=Fs, freq=fc)
p = spec.Periodogram(x, sampling=Fs, window='hamming')
p.run() #Recomputa a psd caso 'x' tenha sido alterado
p.plot()
plt.title("Periodograma (dB) da Spectrum")
plt.tight_layout()
plt.show()
