pred.keys())[0]]['pressure']['function_values'][0, mics[0] - 1, :, 3]) / 2
L = np.shape(pred[list(pred.keys())[0]]['pressure']['function_values'])[2]
fs_pred = L / pred[list(pred.keys())[0]]['pressure']['function_values'][0, 0,
-1, 0]
df = ((fs_pred**-1) * L)**-1
N = ((fs_pred**-1 * df)**-1)
Nfft = np.floor(L / N)
pred_spec = list(
map(
lambda i: fun.msPSD(np.squeeze(pred[list(pred.keys())[0]]['pressure'][
'function_values'][:, :, :, i]).transpose(),
fs_pred,
df=df,
win=False,
ovr=0,
save_fig=False,
plot=False), np.arange(1, 4)))
f_pred = pred_spec[0][0]
Xm_avg_pred = np.array([pred_spec[i][1]
for i in range(len(pred_spec))]).transpose()
Gxx_avg_pred = np.array([pred_spec[i][-1]
for i in range(len(pred_spec))]).transpose()
# f_pred,Gxx_pred,spl_pred = PSD(pred[list(pred.keys())[0]]['pressure']['function_values'][:,:,-1])
f, Xm_inph_pred, Sxx_inph_pred, Gxx_inph_pred, Gxx_avg_inph_pred = fun.msPSD(
np.squeeze(xn_inph_pred),
fs_pred,
df=df,
win=False,
OASPL_pred = 10 * np.log10(np.mean((pred[list(pred.keys())[0]]['pressure']['function_values'][:, :, :, 1:] - np.expand_dims(np.mean(pred[list(pred.keys())[0]]['pressure']['function_values'][:, :, :, 1:], axis = 2), axis = 2)) ** 2, axis=2) / 20e-6 ** 2) #%% p_inph = (pred[list(pred.keys())[0]]['pressure']['function_values'][0, mics[-1] - 1, :, 3] + pred[list(pred.keys())[0]]['pressure']['function_values'][0, mics[0] - 1, :, 3]) / 2 p_outph = (pred[list(pred.keys())[0]]['pressure']['function_values'][0, mics[-1] - 1, :, 3] - pred[list(pred.keys())[0]]['pressure']['function_values'][0, mics[0] - 1, :, 3]) / 2 #%% Nb = 2 L = np.shape(pred[list(pred.keys())[0]]['pressure']['function_values'])[2] fs_pred = L / pred[list(pred.keys())[0]]['pressure']['function_values'][0, 0, -1, 0] df = ((fs_pred**-1)*L)**-1 N = ((fs_pred ** -1 * df)**-1) Nfft = np.floor(L/N) pred_spec = list(map(lambda i: fun.msPSD(np.squeeze(pred[list(pred.keys())[0]]['pressure']['function_values'][:, :, :, i]).transpose(), fs_pred, df = df, win = False, ovr = 0,save_fig = False, plot = False),np.arange(1,4))) f_pred = pred_spec[0][0] Gxx_avg_pred = np.array([pred_spec[i][-1] for i in range(len(pred_spec))]).transpose() f, Xm_inph, Sxx, Gxx, Gxx_avg_inph = fun.msPSD(np.squeeze(p_inph).transpose(), fs_pred, df = df, win = False, ovr = 0, save_fig = False, plot = False) f, Xm, Sxx, Gxx, Gxx_avg_inph_load = fun.msPSD(np.squeeze(p_inph-pred[list(pred.keys())[0]]['pressure']['function_values'][:, :, :, 1]).transpose(), fs_pred, df = df, win = False, ovr = 0, save_fig = False, plot = False) f, Xm_outph, Sxx, Gxx, Gxx_avg_outph = fun.msPSD(np.squeeze(pred[list(pred.keys())[0]]['pressure']['function_values'][:, :, :, 3] - p_inph).transpose(), fs_pred, df = df, win = False, ovr = 0, save_fig = False, plot = False) #%% c = list(mcolors.TABLEAU_COLORS.keys())[:10] #%% Plots predicted pressure time series # Initializes figure with the number of subplots equal to the number of mics specified in the "mics" list fig,ax = plt.subplots(len(mics),1,figsize = (8,6)) # Adds a space in between the subplots and at the bottom for the subplot titles and legend, respectfully. plt.subplots_adjust(hspace = 0.35,bottom = 0.15)
# f1 = lambda a: fun.msPSD(a['Acoustic Data'][:].transpose()/(a['Sensitivities'][:]*1e-3), fs = a['Sampling Rate'][()], df = 5, win= True, ovr = 0.5, plot = False,save_fig=False,f_lim =[10,10e3],levels = [0, 80])
dir ='/Users/danielweitsman/Desktop/Masters_Research/NASA_Coax_2021/Runs/'
caseName = ['BG6','Run151']
leglab = ''
mic =5
df = 5
start_t = 8
end_t = 15
fig,ax = plt.subplots(1,1,figsize = (6.4,4.5))
plt.subplots_adjust(bottom = 0.15)
for i,case in enumerate(caseName):
with h5py.File(os.path.join(dir,caseName[i], 'acs_data.h5'), 'r') as dat_file:
data = (dat_file['Acoustic Data'][:].transpose()/(dat_file['Sensitivities'][:]*1e-3))[int(start_t*dat_file['Sampling Rate'][()]):int(end_t*dat_file['Sampling Rate'][()])]
f,Gxx,Gxx_avg = fun.msPSD(data[:,int(mic-1)], dat_file['Sampling Rate'][()], df=df, ovr=0.5,plot = False,save_fig = False)
ax.plot(f,10*np.log10(Gxx_avg*df/20e-6**2),label = case)
ax.set_xscale('log')
ax.axis([10,20e3,0,80])
ax.set_xlabel('Frequency (Hz)')
ax.set_ylabel('$SPL, \: dB\: (re:\: 20 \: \mu Pa)$')
ax.grid('on')
if isinstance(leglab, list):
ax.legend(leglab)
else:
ax.legend()
ax.set_title('Mic: ' + str(mic))
fs_acs = dat_file['Sampling Rate'][()]
fs_ttl = round((np.mean(np.diff(dat_file['Time (s)'])))**-1)
#%%
# generates time vectors for the tac and acoustic data
t = np.arange(len(ttl)) / fs_ttl
t_acs = np.arange(len(acs_data)) / fs_acs
#%%
LE_ind1, lim_ind1, rpm_nom1, u_rpm1 = fun.rpm_eval(ttl,fs_ttl,start_t,end_t)
ind = list(map(lambda x: bisect(t_acs,x),t[LE_ind1[lim_ind1[0]:lim_ind1[1]]]))
f,fs1,spl,u_low, u_high, Xn_avg,Xm_avg,Xn_avg_filt,Xn_bb = fun.harm_extract(acs_data, ind, fs_acs, 0, harm_filt,filt_shaft_harm=True,Nb = 2)
f_bb,Xm_bb,Sxx_bb,Gxx_bb,Gxx_avg_bb = fun.msPSD(Xn_bb, fs = fs1, df = df, save_fig = False, plot = False)
# construct time vector for the averaged rpm waveform
t_rev = (np.arange(len(Xn_avg))*(rpm_nom1/60)**-1/len(Xn_avg))
#%%
# Initializes figure with the number of subplots equal to the number of mics specified in the "mics" list
fig,ax = plt.subplots(len(mics),1,figsize = (8,6))
# Adds a space in between the subplots and at the bottom for the subplot titles and legend, respectfully.
plt.subplots_adjust(hspace = 0.35,bottom = 0.15)
# Loops through each mic
for i,m in enumerate(mics):
# Computes the mean-square PSD spectra for each mic
f_ms,Xm_ms,Sxx_ms,Gxx_ms,Gxx_avg_ms = fun.msPSD(acs_data[int(fs_acs * start_t):int(fs_acs * end_t),m-1], fs_acs, df=df, ovr=0.75,plot = False,save_fig = False)
# Plots the resulting spectra in dB
bbox_to_anchor=(.25, 0.9))
#%% Overlays the predicted and measured spectra
# Initializes figure with the number of subplots equal to the number of mics specified in the "mics" list
fig, ax = plt.subplots(len(mics), 1, figsize=(8, 6))
# Adds a space in between the subplots and at the bottom for the subplot titles and legend, respectfully.
plt.subplots_adjust(hspace=0.35, bottom=0.15)
# Loops through each mic
for i, m in enumerate(mics):
# Computes the mean-square PSD spectra for each mic
f_exp, Xm_exp, Sxx_exp, Gxx_exp, Gxx_avg_exp = fun.msPSD(
exp[int(fs * start_t):int(fs * end_t), m - 1],
fs,
df=df_exp,
ovr=0.75,
plot=False,
save_fig=False)
# Plots the resulting spectra in dB
if len(mics) > 1:
ax[i].plot(f_exp,
10 * np.log10(Gxx_avg_exp * df_exp / 20e-6**2),
linestyle=linestyle[1])
ax[i].stem(f,
spl[m - 1, :, -1],
linefmt=f'C{1}{linestyle[0]}',
markerfmt=f'C{1}o',
basefmt=f'C{1}')
# ax[i].plot(np.squeeze(pred[list(pred.keys())[0]]['octaveFilterSP/spl_octFilt_spectrum']['function_values'][:, m - 1, :, 0]),np.squeeze(pred[list(pred.keys())[0]]['BPM_total_dB'][:, m - 1, :]))
import pyPostAcsFun as fun
import numpy as np
#%%
mic = 9
# prefix to path containing cases
dir = '//Users/danielweitsman/Box/Jan21Test/dan_thesis/runs/trim_compare/e2/'
# functions to apply to all cases contained in the folder specified as "dir"
f1 = lambda a, b: fun.msPSD(a['Acoustic Data'][:].transpose() /
(a['Sensitivities'][:] * 1e-3),
fs=a['Sampling Rate'][()],
df=5,
win=True,
ovr=0.5,
save_path=b,
save_fig=True,
plot=True,
f_lim=[10, 10e3],
levels=[0, 100])
f2 = lambda a, b: fun.tseries((a['Acoustic Data (mV)'][:].transpose() /
(a['Sensitivities'][:] * 1e-3)),
fs=a['Sampling Rate'][()],
t_lim=[0, 10],
levels=[-0.25, 0.25],
save_path=b,
save_fig=True)
f3 = lambda a, b: fun.spectrogram((a['Acoustic Data'][mic, :].transpose() /
(a['Sensitivities'][mic] * 1e-3)),
fs=a['Sampling Rate'][()],
df=10,
ovr=0.5,
win=True,
Q_filt = lfilter(b, a, dat_file['Motor2 Torque (Nm)'])
Q_avg[i] = np.mean(
dat_file['Motor2 Torque (Nm)'][int(t_min * fs):int(t_max * fs)])
# b, a = butter(4, 2000 / (fs / 2), 'lp')
a_filt = lfilter(b, a, dat_file['Acceleration_Ch2(m_s2)'])
# single sided spectral density of loads
df = .25
f, Gxx, Gxx_avg = fun.msPSD(
dat_file['Motor2 Thrust (Nm)'][int(t_min * fs):int(t_max * fs)],
fs=fs,
df=df,
win=False,
ovr=0,
f_lim=[0, 5e3],
levels=[0, 100],
save_fig=False,
save_path='',
plot=False)
#%%
fig, ax = plt.subplots(1, 1, figsize=(6.4, 4.5))
plt.subplots_adjust(bottom=0.15)
ax.set_title(case + ': ' + str(round(rpm_avg[i], 1)) + ' rpm')
ax.plot(f, 10 * np.log10(Gxx_avg * df))
# ax.set_xscale('log')
ax.set_xlim([f[0], 40])
ax.set_xlabel('Frequency (Hz)')
ax.set_ylabel('$Thrust \ PSD, \ dB \ (re: \ 1 \ N^2/Hz)$')