def test_short_lvm():
data = read('./data/pickle_only.lvm')
np.testing.assert_equal(data[0]['data'][0, 0], 0.914018)
data = read('./data/short.lvm', read_from_pickle=False)
np.testing.assert_equal(data[0]['data'][0, 0], 0.914018)
data = read('./data/short.lvm', read_from_pickle=True)
np.testing.assert_equal(data[0]['data'][0, 0], 0.914018)
data = read('./data/short_new_line_end.lvm',
read_from_pickle=True,
dump_file=False)
np.testing.assert_equal(data[0]['data'][0, 0], 0.914018)
def import_data_lvm(fn):
import lvm_read
data = lvm_read.read(fn)
data = data[0]["data"]
signals = [[i[a] for i in data] for a in [1, 2, 3, 4]]
rot_freq = [i[5] for i in data if i[5]]
return (signals, mean(rot_freq[0:20]))
def plot_real_data_incorrect():
import lvm_read
fn = "./data/teippimittaus_uusi.lvm"
data = lvm_read.read(fn)
channels = (data[0]["Channel names"])
channel_data = data[0]["data"]
channel_names = ['Laser 1', 'Laser 2', 'Laser 3', 'Laser 4']
channel_indexes = [channels.index(name) for name in channel_names]
rounds = [0, 100]
signals = [[i[a] for i in channel_data][1024 * rounds[0]:1024 * rounds[1]]
for a in channel_indexes]
signals = np.array(average(signals))
data = import_data_lvm("matlab/4.00Hz_300mm.lvm")
signals = np.array(average(data[0]))
advanced1 = advanced_roundness_f_coeff(signals, [0, 38, 67, 180])
advanced2 = advanced_roundness_f_coeff(
signals, [0, 37.06875, 65.88984375, 179.41640625])
filtered_advanced1 = filter_fft(advanced1)
filtered_advanced2 = filter_fft(advanced2)
ar1 = get_roundness_profile(advanced1)
ar2 = get_roundness_profile(advanced2)
polar_plot(ar1, "Angles with errors", 0.03)
polar_plot(ar2, "Determined angles", 0.03)
plt.show()
def timing_on_long_short_lvm():
N = 5
tic = time.time()
for i in range(N):
data = read('./data/long.lvm', read_from_pickle=False)
toc = time.time()
print(f'Average time: {(toc-tic)/N:3.1f}s')
def read_lvm(path):
raw_data = lvm_read.read(path)
dataset = pd.DataFrame(
raw_data[0]['data'],
columns=[c.strip() for c in raw_data[0]['Channel names'] if c])
dataset = dataset[[
c for c in dataset.columns if c == 'X_Value' or 'Accel' in c
]]
return dataset.rename(
{
'X_Value': 'Time',
dataset.columns[-1]: 'Acceleration'
},
axis='columns')
fTyp = [("", "*.lvm")]
iDir = os.path.abspath(os.path.dirname(__file__))
tkinter.messagebox.showinfo('lvm形式ファイルための読み込みプログラム',
'処理を実施したいファイルを複数選択してください！')
file = tkinter.filedialog.askopenfilenames(filetypes=fTyp, initialdir=iDir)
##### 選択ファイルリスト作成 #####
list_tmp = list(file)
tkinter.messagebox.showinfo('lvm形式ファイルための読み込みプログラム', list_tmp)
count = 0
for j in range(0, len(list_tmp)):
filepath = list_tmp[j]
basename = os.path.basename(filepath)
lvm_file = read(filepath)
if j == 0:
tmp = lvm_file[0]['data'][0, 0:4] #辞書型からキーを選択してデータを抽出する
u_true = np.sqrt(
2 * tmp[1] * 1000 /
tmp[3]) # U = SQRT (2 x Pd / rho ) [Pa] マノメータの換算には出力電圧を1000倍する
tmp = np.append(tmp, u_true)
ttmp = np.reshape(tmp, (5, 1)).T #抽出後は1次配列となるため，変換して行列にする
df_lvm = pd.DataFrame(ttmp, index=[basename])
else:
tmp = lvm_file[0]['data'][0, 0:4] #辞書型からキーを選択してデータを抽出する
u_true = np.sqrt(2 * tmp[1] * 1000 / tmp[3])
tmp = np.append(tmp, u_true)
import sys, os
my_path = os.path.dirname(os.path.abspath(__file__))
sys.path.insert(0, my_path + '/../')
import FLife
import lvm_read
import numpy as np
import matplotlib.pyplot as plt
import scipy.integrate as si
from scipy.integrate import quad
data = lvm_read.read('./data/m1.lvm')[0]
t = data['data'][:, 0]
x = data['data'][:, 1]
rms = 100
C = 1.8e+22
k = 7.3
Su = 446
x = rms * x / np.std(x)
# Spectral data
sd = FLife.SpectralData(input=x, dt=t[1], nperseg=int(0.1 / t[1]))
#Show the PSD
plt.figure(figsize=(10, 5))
plt.title('PSD')
plt.plot(sd.psd[:, 0], sd.psd[:, 1])
plt.xlim(0, 1200)
plt.show()
from os import path
if __name__== '__main__':
if len(sys.argv) < 2:
print("Filename not given")
sys.exit(-1)
filename = sys.argv[1]
metadata_filename = filename[:-4] + '_metadata.txt'
data_filename = filename[:-4] + '_serialized.txt'
if path.exists(metadata_filename) and path.exists(data_filename):
print("Files are already done")
sys.exit(0)
print(filename)
float_formatter = lambda x: "%.2f" % x
np.set_printoptions(formatter={'float': '{: 0.5f}'.format})
try:
lvm = lvm_read.read(filename, read_from_pickle=False)
except IOError as e:
print(e)
sys.exit(-1)
#print lvm.keys()
lvm_metadata = []
if lvm.has_key('Fs(MHz)'):
lvm_metadata.append(float(lvm['Fs(MHz)']))
else:
#np.insert(lvm_metadata, len(lvm_metadata)-1, 0)
np.asarray(0)
np.savetxt(metadata_filename, np.asarray(lvm_metadata), fmt='%1.4f')
np.savetxt(data_filename, lvm[0]['data'], fmt='%1.4f')
sys.exit(0)
def task(lvm_file, params):
# read params
channel = params['channel']
n_fft = params['n_fft']
win_length = params['win_length']
hop_length = params['hop_length']
win_type = params.get('win_type', 'hann')
threshold = params['threshold']
min_duration = params['min_duration'] / hop_length # convert to frames
print('Processing {}'.format(lvm_file))
# read lvm file
lvm = lvm_read.read(lvm_file)
signal = lvm[0]['data'][:, channel]
# normalize signal
signal -= np.mean(signal)
signal /= np.std(signal)
# process signal
fft = librosa.stft(signal,
n_fft=n_fft,
win_length=win_length,
hop_length=hop_length,
window=win_type)
feats = np.abs(fft)
energy = np.mean(feats, axis=0)
# compute train mask for frames
mask = np.array(energy > threshold, dtype=np.int)
diff = mask[0:-1] ^ mask[1:] # make difference
changes = np.where(diff != 0)[0] # search for changes
changes = np.append(changes, len(mask) - 1) # add the last sample
# convert not train frames to train frames if sequence not train frames length < min duration
start_index = 0
for index in changes:
end_index = index + 1
if mask[start_index] == 0:
duration = end_index - start_index
if duration < min_duration:
mask[start_index:end_index] = 1
start_index = end_index
# compute train mask for signal
signal_mask = np.zeros_like(signal, dtype=np.int)
diff = mask[0:-1] ^ mask[1:] # make difference
changes = np.where(diff != 0)[0] # search for changes
changes = np.append(changes, len(mask) - 1) # add the last sample
start_index = 0
for index in changes:
end_index = index + 1
if mask[start_index] == 1:
# expand signal for one frame left and right
start, end = librosa.frames_to_samples(
[start_index - 1, end_index + 1], hop_length=hop_length)
signal_mask[start:end] = 1
start_index = end_index
results = {
lvm_file: {
'signal': signal,
'signal_mask': signal_mask,
'feats': feats,
'energy': energy,
'mask': mask
}
}
return results
def test_data():
results_ref = {
'Rainflow': 906.217537,
'Rainflow-Goodman': 827.866874,
'Dirlik': 1067.423788,
'Tovo Benasciutti 1': 735.084318,
'Tovo Benasciutti 2': 1114.625812,
'Zhao Baker 1': 985.886435,
'Zhao Baker 2': 1048.549852,
'Narrowband': 711.258072,
'Alpha 0.75': 1086.593252,
'Wirsching Light': 1038.1813918800456,
'Rice': 687.739914,
'Gao Moan': 837.392263,
'Petrucci Zuccarello': 4.322102
}
data = lvm_read.read('./data/m1.lvm')[0]
t = data['data'][:, 0]
x = data['data'][:, 1]
rms = 100
C = 1.8e+22
k = 7.3
Su = 446
x = rms * x / np.std(x)
# Spectral data
sd = FLife.SpectralData(input=x, dt=t[1], nperseg=int(0.1 / t[1]))
# Rainflow reference fatigue life
rf = FLife.Rainflow(sd)
# Spectral methods
dirlik = FLife.Dirlik(sd)
tb = FLife.TovoBenasciutti(sd)
zb = FLife.ZhaoBaker(sd)
nb = FLife.Narrowband(sd)
a075 = FLife.Alpha075(sd)
wl = FLife.WirschingLight(sd)
rice = FLife.Rice(sd)
gm = FLife.GaoMoan(sd)
pz = FLife.PetrucciZuccarello(sd)
# Test PDF's; expected result should be 1
PDFs = {
'Dirlik':
quad(dirlik.get_PDF, a=0, b=np.Inf)[0],
'Rice':
quad(nb.get_PDF, a=0, b=np.Inf)[0],
'Rice -inf':
quad(rice.get_PDF, a=-np.Inf, b=np.Inf)[0],
'Tovo Benasciutti 1':
quad(tb.get_PDF, a=0, b=np.Inf, args=('method 1', ))[0],
'Tovo Benasciutti 2':
quad(tb.get_PDF, a=0, b=np.Inf, args=('method 2', ))[0],
'Zhao Baker 1':
quad(zb.get_PDF, a=0, b=np.Inf, args=('method 1', ))[0],
'Zhao Baker 2':
quad(zb.get_PDF, a=0, b=np.Inf, args=('method 2', ))[0],
}
for method, value in PDFs.items():
np.testing.assert_almost_equal(value,
1.,
decimal=5,
err_msg=f'Method: {method}')
results = {
'Rainflow': rf.get_life(C=C, k=k),
'Rainflow-Goodman': rf.get_life(C=C, k=k, Su=Su),
'Dirlik': dirlik.get_life(C=C, k=k),
'Tovo Benasciutti 1': tb.get_life(C=C, k=k, method='method 1'),
'Tovo Benasciutti 2': tb.get_life(C=C, k=k),
'Zhao Baker 1': zb.get_life(C=C, k=k),
'Zhao Baker 2': zb.get_life(C=C, k=k, method='method 2'),
'Narrowband': nb.get_life(C=C, k=k),
'Alpha 0.75': a075.get_life(C=C, k=k),
'Wirsching Light': wl.get_life(C=C, k=k),
'Rice': rice.get_life(C=C, k=k),
'Gao Moan': gm.get_life(C=C, k=k),
'Petrucci Zuccarello': pz.get_life(C=C, k=k, Su=Su)
}
for method, value in results.items():
if method == 'Petrucci Zuccarello':
compare_to = 'Rainflow-Goodman'
else:
compare_to = 'Rainflow'
err = FLife.tools.relative_error(value, results[compare_to])
print(
f'{method:>19s}:{value:6.0f} s,{100*err:>4.0f} % to {compare_to}')
np.testing.assert_almost_equal(value,
results_ref[method],
decimal=5,
err_msg=f'Method: {method}')
results_via_PDF = {
'Dirlik':
dirlik.get_life(C=C, k=k, integrate_pdf=True),
'Tovo Benasciutti 1':
tb.get_life(C=C, k=k, method='method 1', integrate_pdf=True),
'Tovo Benasciutti 2':
tb.get_life(C=C, k=k, integrate_pdf=True),
'Zhao Baker 1':
zb.get_life(C=C, k=k, integrate_pdf=True),
'Zhao Baker 2':
zb.get_life(C=C, k=k, method='method 2', integrate_pdf=True),
'Narrowband':
nb.get_life(C=C, k=k, integrate_pdf=True),
'Alpha 0.75':
a075.get_life(C=C, k=k),
'Wirsching Light':
wl.get_life(C=C, k=k),
'Rice':
rice.get_life(C=C, k=k),
'Gao Moan':
gm.get_life(C=C, k=k),
'Petrucci Zuccarello':
pz.get_life(C=C, k=k, Su=Su)
}
for method, value in results_via_PDF.items():
np.testing.assert_almost_equal(value / results[method],
1.0,
decimal=2,
err_msg=f'Method: {method}')
import lvm_read
import numpy as np
fname = "./data_demo/LFDI.lvm"
lvm = lvm_read.read(fname)
mat = []
for i in range(lvm['Segments']):
mat.append(np.array(lvm[i]['data'])[:,0])
mat = np.array(mat).T
np.savetxt(fname[0:-3]+"csv", mat[:,0:], delimiter=",")
def test_no_decimal_separator():
data = read('./data/no_decimal_separator.lvm',
read_from_pickle=False,
dump_file=False)
np.testing.assert_equal(data[0]['data'][0, 1], -0.008807)
def test_with_multi_time_column_lvm():
data = read('./data/multi_time_column.lvm',
read_from_pickle=False,
dump_file=False)
np.testing.assert_allclose(data[0]['data'][0],\
np.array([0.000000,-0.035229,0.000000,0.532608]))
def test_with_empty_fields_lvm():
data = read('./data/with_empty_fields.lvm',
read_from_pickle=False,
dump_file=False)
np.testing.assert_equal(data[0]['data'][0, 7], -0.011923)
def test_several_comments():
data = read('./data/with_comments.lvm',
read_from_pickle=False,
dump_file=False)
np.testing.assert_equal(data[0]['data'][0, 1], 1.833787)