def test_lsfd():
f, frf, modal_sim, eta_sim, f0_sim = get_simulated_receptance(
df_Hz=1,
f_start=0,
f_end=5001,
measured_points=8,
show=False,
real_mode=False)
low_lim = 1500
nf = (2 * (len(f) - 1))
while max(prime_factors(nf)) > 5:
f = f[:-1]
frf = frf[:, :-1]
nf = (2 * (len(f) - 1))
df = (f[1] - f[0])
nf = 2 * (len(f) - 1)
ts = 1 / (nf * df) # sampling period
sr = lscf(frf,
low_lim,
6,
ts,
weighing_type='Unity',
reconstruction='LSFD')
fr, xi = complex_freq_to_freq_and_damp(sr[-1])
print('Eigenfrequencies\n', fr)
print('Damping factors\n', xi)
def test_lsce():
from OpenModal.analysis.utility_functions import complex_freq_to_freq_and_damp
""" Test of the Least-Squares Complex Exponential Method """
from OpenModal.analysis.get_simulated_sample import get_simulated_receptance
import matplotlib.pyplot as plt
f, frf, modal_sim, eta_sim, f0_sim = get_simulated_receptance(df_Hz=1,
f_start=0, f_end=5001, measured_points=8, show=False, real_mode=False)
low_lim = 100
nf = (2*(len(f)-low_lim-1))
while max(prime_factors(nf)) > 5:
f = f[:-1]
frf = frf[:, :-1]
nf = (2*(len(f)-low_lim-1))
df = (f[1] - f[0])
nf = 2*(len(f)-low_lim-1)
ts = 1 / (nf * df) # sampling period
n = 12
low_lim = 100
sr = lsce(frf, f[low_lim], low_lim, n, ts, additional_timepoints=0, reconstruction='LSFD')
fr, xi = complex_freq_to_freq_and_damp(sr[-2])
print("fr\n", fr)
print("xi\n", xi)
def test_lsfd():
f, frf, modal_sim, eta_sim, f0_sim = get_simulated_receptance(
df_Hz=1, f_start=0, f_end=5001, measured_points=8, show=False, real_mode=False)
low_lim = 1500
nf = (2*(len(f)-low_lim-1))
while max(prime_factors(nf)) > 5:
f = f[:-1]
frf = frf[:, :-1]
nf = (2*(len(f)-low_lim-1))
df = (f[1] - f[0])
nf = 2*(len(f)-1)
ts = 1 / (nf * df) # sampling period
sr = lscf(frf, low_lim, 6, ts, weighing_type='Unity', reconstruction='LSFD')
fr, xi = complex_freq_to_freq_and_damp(sr[-1])
print('Eigenfrequencies\n', fr)
print('Damping factors\n', xi)
def test_lsce():
from OpenModal.analysis.utility_functions import complex_freq_to_freq_and_damp
""" Test of the Least-Squares Complex Exponential Method """
from OpenModal.analysis.get_simulated_sample import get_simulated_receptance
import matplotlib.pyplot as plt
f, frf, modal_sim, eta_sim, f0_sim = get_simulated_receptance(
df_Hz=1,
f_start=0,
f_end=5001,
measured_points=8,
show=False,
real_mode=False)
low_lim = 100
nf = (2 * (len(f) - low_lim - 1))
while max(prime_factors(nf)) > 5:
f = f[:-1]
frf = frf[:, :-1]
nf = (2 * (len(f) - low_lim - 1))
df = (f[1] - f[0])
nf = 2 * (len(f) - low_lim - 1)
ts = 1 / (nf * df) # sampling period
n = 12
low_lim = 100
sr = lsce(frf,
f[low_lim],
low_lim,
n,
ts,
additional_timepoints=0,
reconstruction='LSFD')
fr, xi = complex_freq_to_freq_and_damp(sr[-2])
print("fr\n", fr)
print("xi\n", xi)
def stabilisation(sr, nmax, err_fn, err_xi):
"""
A function that computes the stabilisation matrices needed for the
stabilisation chart. The computation is focused on comparison of
eigenfrequencies and damping ratios in the present step
(N-th model order) with the previous step ((N-1)-th model order).
:param sr: list of lists of complex natrual frequencies
:param n: maximum number of degrees of freedom
:param err_fn: relative error in frequency
:param err_xi: relative error in damping
:return fn_temp eigenfrequencies matrix
:return xi_temp: updated damping matrix
:return test_fn: updated eigenfrequencies stabilisation test matrix
:return test_xi: updated damping stabilisation test matrix
@author: Blaz Starc
@contact: [email protected]
"""
# TODO: check this later for optimisation # this doffers by LSCE and LSCF
fn_temp = np.zeros((2*nmax, nmax), dtype = 'double')
xi_temp = np.zeros((2*nmax, nmax), dtype = 'double')
test_fn = np.zeros((2*nmax, nmax), dtype = 'int')
test_xi = np.zeros((2*nmax, nmax), dtype = 'int')
for nr, n in enumerate(range(nmax)):
fn, xi = complex_freq_to_freq_and_damp(sr[nr])
fn, xi = redundant_values(fn, xi, 1e-3) # elimination of conjugate values in
# order to decrease computation time
if n == 1:
# first step
fn_temp[0:len(fn), 0:1] = fn
xi_temp[0:len(fn), 0:1] = xi
else:
# Matrix test is created for comparison between present(N-th) and
# previous (N-1-th) data (eigenfrequencies). If the value equals:
# --> 1, the data is within relative tolerance err_fn
# --> 0, the data is outside the relative tolerance err_fn
fn_test = np.zeros((len(fn), len(fn_temp[:, n - 1])), dtype ='int')
for i in range(0, len(fn)):
for j in range(0, len(fn_temp[0:2*(n), n-1])):
if fn_temp[j, n-2] == 0:
fn_test[i,j] = 0
else:
if np.abs((fn[i] - fn_temp[j, n-2])/fn_temp[j, n-2]) < err_fn:
fn_test[i,j] = 1
else: fn_test[i,j] = 0
for i in range(0, len(fn)):
test_fn[i, n - 1] = np.sum(fn_test[i, :]) # all rows are summed together
# The same procedure as for eigenfrequencies is applied for damping
xi_test = np.zeros((len(xi), len(xi_temp[:, n - 1])), dtype ='int')
for i in range(0, len(xi)):
for j in range(0, len(xi_temp[0:2*(n), n-1])):
if xi_temp[j, n-2]==0:
xi_test[i,j] = 0
else:
if np.abs((xi[i] - xi_temp[j, n-2])/xi_temp[j, n-2]) < err_xi:
xi_test[i,j] = 1
else: xi_test[i,j] = 0
for i in range(0, len(xi)):
test_xi[i, n - 1] = np.sum(xi_test[i, :])
# If the frequency/damping values corresponded to the previous iteration,
# a mean of the two values is computed, otherwise the value stays the same
for i in range(0, len(fn)):
for j in range(0, len(fn_temp[0:2*(n), n-1])):
if fn_test[i,j] == 1:
fn_temp[i, n - 1] = (fn[i] + fn_temp[j, n - 2]) / 2
elif fn_test[i,j] == 0:
fn_temp[i, n - 1] = fn[i]
for i in range(0, len(fn)):
for j in range(0, len(fn_temp[0:2*(n), n-1])):
if xi_test[i,j] == 1:
xi_temp[i, n - 1] = (xi[i] + xi_temp[j, n - 2]) / 2
elif xi_test[i,j] == 0:
xi_temp[i, n - 1] = xi[i]
return fn_temp, xi_temp, test_fn, test_xi
def stabilisation(sr, nmax, err_fn, err_xi):
"""
A function that computes the stabilisation matrices needed for the
stabilisation chart. The computation is focused on comparison of
eigenfrequencies and damping ratios in the present step
(N-th model order) with the previous step ((N-1)-th model order).
:param sr: list of lists of complex natrual frequencies
:param n: maximum number of degrees of freedom
:param err_fn: relative error in frequency
:param err_xi: relative error in damping
:return fn_temp eigenfrequencies matrix
:return xi_temp: updated damping matrix
:return test_fn: updated eigenfrequencies stabilisation test matrix
:return test_xi: updated damping stabilisation test matrix
@author: Blaz Starc
@contact: [email protected]
"""
# TODO: check this later for optimisation # this doffers by LSCE and LSCF
fn_temp = np.zeros((2*nmax, nmax), dtype = 'double')
xi_temp = np.zeros((2*nmax, nmax), dtype = 'double')
test_fn = np.zeros((2*nmax, nmax), dtype = 'int')
test_xi = np.zeros((2*nmax, nmax), dtype = 'int')
for nr, n in enumerate(range(nmax)):
fn, xi = complex_freq_to_freq_and_damp(sr[nr])
fn, xi = redundant_values(fn, xi, 1e-3) # elimination of conjugate values in
# order to decrease computation time
if n == 1:
# first step
fn_temp[0:len(fn), 0:1] = fn
xi_temp[0:len(fn), 0:1] = xi
else:
# Matrix test is created for comparison between present(N-th) and
# previous (N-1-th) data (eigenfrequencies). If the value equals:
# --> 1, the data is within relative tolerance err_fn
# --> 0, the data is outside the relative tolerance err_fn
fn_test = np.zeros((len(fn), len(fn_temp[:, n - 1])), dtype ='int')
for i in range(0, len(fn)):
for j in range(0, len(fn_temp[0:2*(n), n-1])):
if fn_temp[j, n-2] == 0:
fn_test[i,j] = 0
else:
if np.abs((fn[i] - fn_temp[j, n-2])/fn_temp[j, n-2]) < err_fn:
fn_test[i,j] = 1
else: fn_test[i,j] = 0
for i in range(0, len(fn)):
test_fn[i, n - 1] = np.sum(fn_test[i, :]) # all rows are summed together
# The same procedure as for eigenfrequencies is applied for damping
xi_test = np.zeros((len(xi), len(xi_temp[:, n - 1])), dtype ='int')
for i in range(0, len(xi)):
for j in range(0, len(xi_temp[0:2*(n), n-1])):
if xi_temp[j, n-2]==0:
xi_test[i,j] = 0
else:
if np.abs((xi[i] - xi_temp[j, n-2])/xi_temp[j, n-2]) < err_xi:
xi_test[i,j] = 1
else: xi_test[i,j] = 0
for i in range(0, len(xi)):
test_xi[i, n - 1] = np.sum(xi_test[i, :])
# If the frequency/damping values corresponded to the previous iteration,
# a mean of the two values is computed, otherwise the value stays the same
for i in range(0, len(fn)):
for j in range(0, len(fn_temp[0:2*(n), n-1])):
if fn_test[i,j] == 1:
fn_temp[i, n - 1] = (fn[i] + fn_temp[j, n - 2]) / 2
elif fn_test[i,j] == 0:
fn_temp[i, n - 1] = fn[i]
for i in range(0, len(fn)):
for j in range(0, len(fn_temp[0:2*(n), n-1])):
if xi_test[i,j] == 1:
xi_temp[i, n - 1] = (xi[i] + xi_temp[j, n - 2]) / 2
elif xi_test[i,j] == 0:
xi_temp[i, n - 1] = xi[i]
return fn_temp, xi_temp, test_fn, test_xi
