def calc_loc_errors(tdoa_errors_std, m, sound_speed_mps, hydrophones_config,
hydrophone_pairs):
""" Calculates localization errors. Eq. (8) in Mouy et al. 2018."""
A = defineJacobian(hydrophones_config, m, sound_speed_mps,
hydrophone_pairs)
Cm = (tdoa_errors_std**2) * np.linalg.inv(np.dot(
A.transpose(), A)) # Model covariance matrix for IID
err_std = np.sqrt(np.diag(Cm))
return pd.DataFrame({
'x_std': [err_std[0]],
'y_std': [err_std[1]],
'z_std': [err_std[2]]
})
os.mkdir(outdir)
# Define receiver pairs for TDOAs
Rpairs = loclib.defineReceiverPairs(nReceivers)
# Repeats optimization nIter times to ensure stability
for i in range(nIter):
# Closes all open figures
plt.close("all")
# Optimize array configuration
R, Rchanges, acceptRateChanges, Cost, processingTime = loclib.optimizeArray(
ReceiverBounds, nReceivers, AnnealingSchedule, S, Rpairs, V,
NoiseVariance)
# Get list of Jacobian matrice for each source
J2 = loclib.defineJacobian(R, S, V, Rpairs)
# Calculates localization uncertainty for each source
Uncertainties2 = loclib.getUncertainties(J2, NoiseVariance)
# Plots unceratinties of optimized array
loclib.plotArrayUncertainties(R, S, Uncertainties2)
plt.savefig(os.path.join(
outdir, 'UncertaintiesPlot' + '_iteration-' + str(i + 1) + '.png'),
bbox_inches='tight')
# Plots Optimization results
loclib.plotOptimizationResults(outdir, nReceivers, Rchanges, Cost,
acceptRateChanges, R, i)
# Save paraneters and results to pickle file
outfilename = os
data = {
"outroot": outroot,
def getArrayUncertainties(R, radius, spacing, V, NoiseSTD, contoursValues):
# Virtual sources coordinates -> Cube of points (Cartesian coordinates)
vec = np.arange(-radius, radius + spacing, spacing)
X, Y, Z = np.meshgrid(vec, vec, vec, indexing='ij')
Sx = np.reshape(X, X.shape[0] * X.shape[1] * X.shape[2])
Sy = np.reshape(Y, Y.shape[0] * Y.shape[1] * Y.shape[2])
Sz = np.reshape(Z, Z.shape[0] * Z.shape[1] * Z.shape[2])
S = pd.DataFrame({'x': Sx, 'y': Sy, 'z': Sz})
# find location of slice
ind = np.argmin(abs(vec))
sliceValue = vec[ind]
# Nb of receivers
nReceivers = R.shape[0]
# Variance of TDOA measurement errors
NoiseVariance = NoiseSTD**2
# Define receiver pairs for TDOAs
Rpairs = loclib.defineReceiverPairs(nReceivers)
# Get list of Jacobian matrice for each source
J = loclib.defineJacobian(R, S, V, Rpairs)
# Calculates localization uncertainty for each source
Uncertainties = loclib.getUncertainties(J, NoiseVariance)
# Plots unceratinties of optimized array
loclib.plotArrayUncertainties(R, S, Uncertainties)
# PLot hydrophone locations
f0 = plt.figure()
ax0 = f0.add_subplot(111, projection='3d')
ax0.scatter(R['x'], R['y'], R['z'], s=30, c='black')
ax0.set_xlabel('X (m)', labelpad=10)
ax0.set_ylabel('Y (m)', labelpad=10)
ax0.set_zlabel('Z (m)', labelpad=10)
plt.show()
# Define and plot plane slices
f, (ax1, ax2, ax3) = plt.subplots(1, 3, sharey=False, figsize=(16, 5))
## XY plane
XY = np.zeros([len(vec), len(vec)])
for i in range(len(vec)):
for jj in range(len(vec)):
idx = S.index[(S['x'] == vec[i]) & (S['y'] == vec[jj]) &
(S['z'] == sliceValue)][0]
XY[i, jj] = Uncertainties['rms'][idx]
CS_XY = ax1.contour(vec, vec, XY, levels=contoursValues, colors=['k'])
# Receivers
ax1.plot(R['x'], R['y'], 'go')
ax1.set_xlabel('X(m)')
ax1.set_ylabel('Y(m)')
ax1.grid(True)
im = ax1.imshow(XY,
interpolation='bilinear',
origin='lower',
cmap=cm.jet,
extent=(-radius, radius, -radius, radius),
norm=colors.Normalize(vmin=0, vmax=10))
ax1.set_aspect('auto')
cbar = f.colorbar(im, ax=ax1)
cbar.ax.set_ylabel('Uncertainty (m)')
## XZ plane
XZ = np.zeros([len(vec), len(vec)])
for i in range(len(vec)):
for jj in range(len(vec)):
idx = S.index[(S['x'] == vec[i]) & (S['z'] == vec[jj]) &
(S['y'] == sliceValue)][0]
XZ[i, jj] = Uncertainties['rms'][idx]
CS_XZ = ax2.contour(vec, vec, XZ, levels=contoursValues, colors=['k'])
# Receivers
ax2.plot(R['x'], R['z'], 'go')
ax2.set_xlabel('X(m)')
ax2.set_ylabel('Z(m)')
ax2.grid(True)
im = ax2.imshow(XZ,
interpolation='bilinear',
origin='lower',
cmap=cm.jet,
extent=(-radius, radius, -radius, radius),
norm=colors.Normalize(vmin=0, vmax=10))
ax2.set_aspect('auto')
cbar = f.colorbar(im, ax=ax2)
cbar.ax.set_ylabel('Uncertainty (m)')
## YZ plane
YZ = np.zeros([len(vec), len(vec)])
for i in range(len(vec)):
for jj in range(len(vec)):
idx = S.index[(S['y'] == vec[i]) & (S['z'] == vec[jj]) &
(S['x'] == sliceValue)][0]
YZ[i, jj] = Uncertainties['rms'][idx]
CS_YZ = ax3.contour(vec, vec, YZ, levels=contoursValues, colors=['k'])
# Receivers
ax3.plot(R['y'], R['z'], 'go')
ax3.set_xlabel('Y(m)')
ax3.set_ylabel('Z(m)')
ax3.grid(True)
im = ax3.imshow(YZ,
interpolation='bilinear',
origin='lower',
cmap='jet',
extent=(-radius, radius, -radius, radius),
norm=colors.Normalize(vmin=0, vmax=10))
cbar = f.colorbar(im, ax=ax3)
cbar.ax.set_ylabel('Uncertainty (m)')
# from mpl_toolkits.axes_grid1 import make_axes_locatable
# divider = make_axes_locatable(plt.gca())
# cax = divider.append_axes("right", "5%", pad="3%")
# plt.colorbar(im, cax=cax)
#plt.colorbar(im,ax=ax3)
ax3.set_aspect('auto')
#plt.tight_layout()
plt.show()
#plt.tight_layout()
# Extract contours lines
coord_CS_XY = []
coord_CS_XZ = []
coord_CS_YZ = []
for i in range(len(contoursValues)):
p1 = CS_XY.collections[i].get_paths()[0]
coord_CS_XY.append(p1.vertices)
p2 = CS_XZ.collections[i].get_paths()[0]
coord_CS_XZ.append(p2.vertices)
p3 = CS_YZ.collections[i].get_paths()[0]
coord_CS_YZ.append(p3.vertices)
return coord_CS_XY, coord_CS_XZ, coord_CS_YZ
