def _extract_result_ppci_to_pp(net, ppc, ppci):
# convert to pandapower indices
ppc = _copy_results_ppci_to_ppc(ppci, ppc, mode="se")
# inits empty result tables
init_results(net, mode="se")
# writes res_bus.vm_pu / va_degree and branch res
_extract_results_se(net, ppc)
# additionally, write bus power demand results (these are not written in _extract_results)
mapping_table = net["_pd2ppc_lookups"]["bus"]
net.res_bus_est.index = net.bus.index
net.res_bus_est.p_mw = get_values(ppc["bus"][:, 2], net.bus.index.values,
mapping_table)
net.res_bus_est.q_mvar = get_values(ppc["bus"][:, 3], net.bus.index.values,
mapping_table)
return net
def _extract_result_ppci_to_pp(net, ppc, ppci):
# convert to pandapower indices
ppc = _copy_results_ppci_to_ppc(ppci, ppc, mode="se")
# extract results from ppc
try:
_add_pf_options(net,
tolerance_mva=1e-8,
trafo_loading="current",
numba=True,
ac=True,
algorithm='nr',
max_iteration="auto")
except:
pass
# writes res_bus.vm_pu / va_degree and res_line
_extract_results_se(net, ppc)
# restore backup of previous results
_rename_results(net)
# additionally, write bus power demand results (these are not written in _extract_results)
mapping_table = net["_pd2ppc_lookups"]["bus"]
net.res_bus_est.index = net.bus.index
net.res_bus_est.p_mw = get_values(ppc["bus"][:, 2], net.bus.index.values,
mapping_table)
net.res_bus_est.q_mvar = get_values(ppc["bus"][:, 3], net.bus.index.values,
mapping_table)
_clean_up(net)
# delete results which are not correctly calculated
for k in list(net.keys()):
if k.startswith("res_") and k.endswith("_est") and \
k not in ("res_bus_est", "res_line_est", "res_trafo_est", "res_trafo3w_est"):
del net[k]
return net
def estimate(self,
v_start=None,
delta_start=None,
calculate_voltage_angles=True):
"""
The function estimate is the main function of the module. It takes up to three input
arguments: v_start, delta_start and calculate_voltage_angles. The first two are the initial
state variables for the estimation process. Usually they can be initialized in a
"flat-start" condition: All voltages being 1.0 pu and all voltage angles being 0 degrees.
In this case, the parameters can be left at their default values (None). If the estimation
is applied continuously, using the results from the last estimation as the starting
condition for the current estimation can decrease the amount of iterations needed to
estimate the current state. The third parameter defines whether all voltage angles are
calculated absolutely, including phase shifts from transformers. If only the relative
differences between buses are required, this parameter can be set to False. Returned is a
boolean value, which is true after a successful estimation and false otherwise.
The resulting complex voltage will be written into the pandapower network. The result
fields are found res_bus_est of the pandapower network.
INPUT:
**net** - The net within this line should be created
**v_start** (np.array, shape=(1,), optional) - Vector with initial values for all
voltage magnitudes in p.u. (sorted by bus index)
**delta_start** (np.array, shape=(1,), optional) - Vector with initial values for all
voltage angles in degrees (sorted by bus index)
OPTIONAL:
**calculate_voltage_angles** - (bool) - Take into account absolute voltage angles and
phase shifts in transformers Default is True.
OUTPUT:
**successful** (boolean) - True if the estimation process was successful
Optional estimation variables:
The bus power injections can be accessed with *se.s_node_powers* and the estimated
values corresponding to the (noisy) measurement values with *se.hx*. (*hx* denotes h(x))
EXAMPLE:
success = estimate(np.array([1.0, 1.0, 1.0]), np.array([0.0, 0.0, 0.0]))
"""
if self.net is None:
raise UserWarning("Component was not initialized with a network.")
t0 = time()
# add initial values for V and delta
# node voltages
# V<delta
if v_start is None:
v_start = np.ones(self.net.bus.shape[0])
if delta_start is None:
delta_start = np.zeros(self.net.bus.shape[0])
# initialize result tables if not existent
_copy_power_flow_results(self.net)
# initialize ppc
ppc, ppci = _init_ppc(self.net, v_start, delta_start,
calculate_voltage_angles)
# add measurements to ppci structure
ppci = _add_measurements_to_ppc(self.net, ppci, self.s_ref)
# calculate relevant vectors from ppci measurements
z, self.pp_meas_indices, r_cov = _build_measurement_vectors(ppci)
# number of nodes
n_active = len(np.where(ppci["bus"][:, 1] != 4)[0])
slack_buses = np.where(ppci["bus"][:, 1] == 3)[0]
# Check if observability criterion is fulfilled and the state estimation is possible
if len(z) < 2 * n_active - 1:
self.logger.error("System is not observable (cancelling)")
self.logger.error(
"Measurements available: %d. Measurements required: %d" %
(len(z), 2 * n_active - 1))
return False
# set the starting values for all active buses
v_m = ppci["bus"][:, 7]
delta = ppci["bus"][:, 8] * np.pi / 180 # convert to rad
delta_masked = np.ma.array(delta, mask=False)
delta_masked.mask[slack_buses] = True
non_slack_buses = np.arange(len(delta))[~delta_masked.mask]
# matrix calculation object
sem = wls_matrix_ops(ppci, slack_buses, non_slack_buses, self.s_ref)
# state vector
E = np.concatenate((delta_masked.compressed(), v_m))
# invert covariance matrix
r_inv = csr_matrix(np.linalg.inv(np.diagflat(r_cov)**2))
current_error = 100.
cur_it = 0
G_m, r, H, h_x = None, None, None, None
while current_error > self.tolerance and cur_it < self.max_iterations:
self.logger.debug(" Starting iteration %d" % (1 + cur_it))
try:
# create h(x) for the current iteration
h_x = sem.create_hx(v_m, delta)
# residual r
r = csr_matrix(z - h_x).T
# jacobian matrix H
H = csr_matrix(sem.create_jacobian(v_m, delta))
# gain matrix G_m
# G_m = H^t * R^-1 * H
G_m = H.T * (r_inv * H)
# state vector difference d_E
# d_E = G_m^-1 * (H' * R^-1 * r)
d_E = spsolve(G_m, H.T * (r_inv * r))
E += d_E
# update V/delta
delta[non_slack_buses] = E[:len(non_slack_buses)]
v_m = np.squeeze(E[len(non_slack_buses):])
# prepare next iteration
cur_it += 1
current_error = np.max(np.abs(d_E))
self.logger.debug("Current error: %.7f" % current_error)
except np.linalg.linalg.LinAlgError:
self.logger.error(
"A problem appeared while using the linear algebra methods."
"Check and change the measurement set.")
return False
# print output for results
if current_error <= self.tolerance:
successful = True
self.logger.debug(
"WLS State Estimation successful (%d iterations)" % cur_it)
else:
successful = False
self.logger.debug(
"WLS State Estimation not successful (%d/%d iterations)" %
(cur_it, self.max_iterations))
# store results for all elements
# calculate bus power injections
v_cpx = v_m * np.exp(1j * delta)
bus_powers_conj = np.zeros(len(v_cpx), dtype=np.complex128)
for i in range(len(v_cpx)):
bus_powers_conj[i] = np.dot(sem.Y_bus[i, :], v_cpx) * np.conjugate(
v_cpx[i])
ppci["bus"][:, 2] = bus_powers_conj.real # saved in per unit
ppci["bus"][:, 3] = -bus_powers_conj.imag # saved in per unit
ppci["bus"][:, 7] = v_m
ppci["bus"][:, 8] = delta * 180 / np.pi # convert to degree
# calculate line results (in ppc_i)
s_ref, bus, gen, branch = _get_pf_variables_from_ppci(ppci)[0:4]
out = np.flatnonzero(branch[:,
BR_STATUS] == 0) # out-of-service branches
br = np.flatnonzero(branch[:, BR_STATUS]).astype(
int) # in-service branches
# complex power at "from" bus
Sf = v_cpx[np.real(branch[br, F_BUS]).astype(int)] * np.conj(
sem.Yf[br, :] * v_cpx) * s_ref
# complex power injected at "to" bus
St = v_cpx[np.real(branch[br, T_BUS]).astype(int)] * np.conj(
sem.Yt[br, :] * v_cpx) * s_ref
branch[np.ix_(br, [PF, QF, PT, QT])] = np.c_[Sf.real, Sf.imag, St.real,
St.imag]
branch[np.ix_(out, [PF, QF, PT, QT])] = np.zeros((len(out), 4))
et = time() - t0
ppci = _store_results_from_pf_in_ppci(ppci, bus, gen, branch,
successful, cur_it, et)
# convert to pandapower indices
ppc = _copy_results_ppci_to_ppc(ppci, ppc, mode="se")
# extract results from ppc
_add_pf_options(self.net,
tolerance_kva=1e-5,
trafo_loading="current",
numba=True,
ac=True,
algorithm='nr',
max_iteration="auto")
# writes res_bus.vm_pu / va_degree and res_line
_extract_results_se(self.net, ppc)
# restore backup of previous results
_rename_results(self.net)
# additionally, write bus power injection results (these are not written in _extract_results)
mapping_table = self.net["_pd2ppc_lookups"]["bus"]
self.net.res_bus_est.p_kw = -get_values(
ppc["bus"][:, 2], self.net.bus.index.values,
mapping_table) * self.s_ref / 1e3
self.net.res_bus_est.q_kvar = -get_values(
ppc["bus"][:, 3], self.net.bus.index.values,
mapping_table) * self.s_ref / 1e3
# store variables required for chi^2 and r_N_max test:
self.R_inv = r_inv.toarray()
self.Gm = G_m.toarray()
self.r = r.toarray()
self.H = H.toarray()
self.Ht = self.H.T
self.hx = h_x
self.V = v_m
self.delta = delta
# delete results which are not correctly calculated
for k in list(self.net.keys()):
if k.startswith("res_") and k.endswith("_est") and \
k not in ("res_bus_est", "res_line_est", "res_trafo_est", "res_trafo3w_est"):
del self.net[k]
return successful
def estimate(self,
v_start='flat',
delta_start='flat',
calculate_voltage_angles=True,
zero_injection=None,
fuse_buses_with_bb_switch='all'):
"""
The function estimate is the main function of the module. It takes up to three input
arguments: v_start, delta_start and calculate_voltage_angles. The first two are the initial
state variables for the estimation process. Usually they can be initialized in a
"flat-start" condition: All voltages being 1.0 pu and all voltage angles being 0 degrees.
In this case, the parameters can be left at their default values (None). If the estimation
is applied continuously, using the results from the last estimation as the starting
condition for the current estimation can decrease the amount of iterations needed to
estimate the current state. The third parameter defines whether all voltage angles are
calculated absolutely, including phase shifts from transformers. If only the relative
differences between buses are required, this parameter can be set to False. Returned is a
boolean value, which is true after a successful estimation and false otherwise.
The resulting complex voltage will be written into the pandapower network. The result
fields are found res_bus_est of the pandapower network.
INPUT:
**net** - The net within this line should be created
**v_start** (np.array, shape=(1,), optional) - Vector with initial values for all
voltage magnitudes in p.u. (sorted by bus index)
**delta_start** (np.array, shape=(1,), optional) - Vector with initial values for all
voltage angles in degrees (sorted by bus index)
OPTIONAL:
**calculate_voltage_angles** - (bool) - Take into account absolute voltage angles and
phase shifts in transformers Default is True
**zero_injection** - (str, iterable, None) - Defines which buses are zero injection bus or the method
to identify zero injection bus, with 'wls_estimator' virtual measurements will be added, with
'wls_estimator with zero constraints' the buses will be handled as constraints
"auto": all bus without p,q measurement, without p, q value (load, sgen...) and aux buses will be
identified as zero injection bus
"aux_bus": only aux bus will be identified as zero injection bus
None: no bus will be identified as zero injection bus
iterable: the iterable should contain index of the zero injection bus and also aux bus will be identified
as zero-injection bus
**fuse_buses_with_bb_switch** - (str, iterable, None) - Defines how buses with closed bb switches should
be handled, if fuse buses will only fused to one for calculation, if not fuse, an auxiliary bus and
auxiliary line will be automatically added to the network to make the buses with different p,q injection
measurements identifieble
"all": all buses with bb-switches will be fused, the same as the default behaviour in load flow
None: buses with bb-switches and individual p,q measurements will be reconfigurated
by auxiliary elements
iterable: the iterable should contain the index of buses to be fused, the behaviour is contigous e.g.
if one of the bus among the buses connected through bb switch is given, then all of them will still
be fused
OUTPUT:
**successful** (boolean) - True if the estimation process was successful
Optional estimation variables:
The bus power injections can be accessed with *se.s_node_powers* and the estimated
values corresponding to the (noisy) measurement values with *se.hx*. (*hx* denotes h(x))
EXAMPLE:
success = estimate(np.array([1.0, 1.0, 1.0]), np.array([0.0, 0.0, 0.0]))
"""
if self.net is None:
raise UserWarning("Component was not initialized with a network.")
t0 = time()
# change the configuration of the pp net to avoid auto fusing of buses connected
# through bb switch with elements on each bus if this feature enabled
bus_to_be_fused = None
if fuse_buses_with_bb_switch != 'all' and not self.net.switch.empty:
if isinstance(fuse_buses_with_bb_switch, str):
raise UserWarning(
"fuse_buses_with_bb_switch parameter is not correctly initialized"
)
elif hasattr(fuse_buses_with_bb_switch, '__iter__'):
bus_to_be_fused = fuse_buses_with_bb_switch
_add_aux_elements_for_bb_switch(self.net, bus_to_be_fused)
# add initial values for V and delta
# node voltages
# V<delta
if v_start is None:
v_start = "flat"
if delta_start is None:
delta_start = "flat"
# initialize result tables if not existent
_copy_power_flow_results(self.net)
# initialize ppc
ppc, ppci = _init_ppc(self.net, v_start, delta_start,
calculate_voltage_angles)
# add measurements to ppci structure
ppci = _add_measurements_to_ppc(self.net, ppci, zero_injection)
# Finished converting pandapower network to ppci
# Estimate voltage magnitude and angle with the given estimator
delta, v_m = self.estimator.estimate(ppci)
# store results for all elements
# calculate bus power injections
v_cpx = v_m * np.exp(1j * delta)
bus_powers_conj = np.zeros(len(v_cpx), dtype=np.complex128)
for i in range(len(v_cpx)):
bus_powers_conj[i] = np.dot(ppci['internal']['Y_bus'][i, :],
v_cpx) * np.conjugate(v_cpx[i])
ppci["bus"][:, 2] = bus_powers_conj.real # saved in per unit
ppci["bus"][:, 3] = -bus_powers_conj.imag # saved in per unit
ppci["bus"][:, 7] = v_m
ppci["bus"][:, 8] = delta * 180 / np.pi # convert to degree
# calculate line results (in ppc_i)
s_ref, bus, gen, branch = _get_pf_variables_from_ppci(ppci)[0:4]
out = np.flatnonzero(branch[:,
BR_STATUS] == 0) # out-of-service branches
br = np.flatnonzero(branch[:, BR_STATUS]).astype(
int) # in-service branches
# complex power at "from" bus
Sf = v_cpx[np.real(branch[br, F_BUS]).astype(int)] * np.conj(
ppci['internal']['Yf'][br, :] * v_cpx) * s_ref
# complex power injected at "to" bus
St = v_cpx[np.real(branch[br, T_BUS]).astype(int)] * np.conj(
ppci['internal']['Yt'][br, :] * v_cpx) * s_ref
branch[np.ix_(br, [PF, QF, PT, QT])] = np.c_[Sf.real, Sf.imag, St.real,
St.imag]
branch[np.ix_(out, [PF, QF, PT, QT])] = np.zeros((len(out), 4))
et = time() - t0
ppci = _store_results_from_pf_in_ppci(ppci, bus, gen, branch,
self.estimator.successful,
self.estimator.iterations, et)
# convert to pandapower indices
ppc = _copy_results_ppci_to_ppc(ppci, ppc, mode="se")
# extract results from ppc
_add_pf_options(self.net,
tolerance_mva=1e-8,
trafo_loading="current",
numba=True,
ac=True,
algorithm='nr',
max_iteration="auto")
# writes res_bus.vm_pu / va_degree and res_line
_extract_results_se(self.net, ppc)
# restore backup of previous results
_rename_results(self.net)
# additionally, write bus power injection results (these are not written in _extract_results)
mapping_table = self.net["_pd2ppc_lookups"]["bus"]
self.net.res_bus_est.p_mw = -get_values(
ppc["bus"][:, 2], self.net.bus.index.values, mapping_table)
self.net.res_bus_est.q_mvar = -get_values(
ppc["bus"][:, 3], self.net.bus.index.values, mapping_table)
_clean_up(self.net)
# clear the aux elements and calculation results created for the substitution of bb switches
if fuse_buses_with_bb_switch != 'all' and not self.net.switch.empty:
_drop_aux_elements_for_bb_switch(self.net)
# delete results which are not correctly calculated
for k in list(self.net.keys()):
if k.startswith("res_") and k.endswith("_est") and \
k not in ("res_bus_est", "res_line_est", "res_trafo_est", "res_trafo3w_est"):
del self.net[k]
return self.estimator.successful