Pythonfiles
This commit is contained in:
@@ -19,7 +19,8 @@
|
||||
"import Stochastisches_Modell\n",
|
||||
"from Stochastisches_Modell import StochastischesModell\n",
|
||||
"import Export\n",
|
||||
"import Netzqualität_Genauigkeit"
|
||||
"import Netzqualität_Genauigkeit\n",
|
||||
"import Datumsfestlegung"
|
||||
],
|
||||
"outputs": [],
|
||||
"execution_count": null
|
||||
@@ -418,6 +419,24 @@
|
||||
"id": "3b38998a2765b74d",
|
||||
"outputs": [],
|
||||
"execution_count": null
|
||||
},
|
||||
{
|
||||
"metadata": {},
|
||||
"cell_type": "code",
|
||||
"outputs": [],
|
||||
"execution_count": null,
|
||||
"source": [
|
||||
"# Datumsfestlegung: Bitte geben Sie nachfolgend die Koordinatenkomponenten an, die das Datum definieren sollen\n",
|
||||
"\n",
|
||||
"auswahl = [\n",
|
||||
" (\"101\",\"X\"), (\"101\",\"Y\"), # Punkt 101 nur Lage\n",
|
||||
" (\"205\",\"X\"), (\"205\",\"Y\"), (\"205\",\"Z\"), # Punkt 205 voll\n",
|
||||
" (\"330\",\"Z\") # Punkt 330 nur Höhe\n",
|
||||
"]\n",
|
||||
"\n",
|
||||
"aktive_unbekannte_indices = Datumsfestlegung.datumskomponenten(auswahl, liste_punktnummern)"
|
||||
],
|
||||
"id": "c37670a07848d977"
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
|
||||
@@ -1,81 +1,154 @@
|
||||
import sympy as sp
|
||||
from typing import List, Iterable, Tuple
|
||||
from typing import Iterable, List, Sequence, Tuple, Optional
|
||||
|
||||
def raenderungsmatrix_G(
|
||||
x0: sp.Matrix,
|
||||
idx_X: List[int],
|
||||
idx_Y: List[int],
|
||||
idx_Z: List[int],
|
||||
mit_massstab: bool = True,
|
||||
|
||||
class Datumsfestlegung:
|
||||
|
||||
@staticmethod
|
||||
def datumskomponenten(
|
||||
auswahl: Iterable[Tuple[str, str]],
|
||||
liste_punktnummern: Sequence[str],
|
||||
*,
|
||||
layout: str = "XYZ"
|
||||
) -> List[int]:
|
||||
punkt2pos = {str(p): i for i, p in enumerate(liste_punktnummern)}
|
||||
|
||||
layout = layout.upper()
|
||||
if layout != "XYZ":
|
||||
raise ValueError("Nur layout='XYZ' unterstützt (wie bei euch).")
|
||||
comp2off = {"X": 0, "Y": 1, "Z": 2}
|
||||
|
||||
aktive: List[int] = []
|
||||
for pt, comp in auswahl:
|
||||
spt = str(pt)
|
||||
c = comp.upper()
|
||||
if spt not in punkt2pos:
|
||||
raise KeyError(f"Punkt '{pt}' nicht in liste_punktnummern.")
|
||||
if c not in comp2off:
|
||||
raise ValueError(f"Komponente '{comp}' ungültig. Nur X,Y,Z.")
|
||||
p = punkt2pos[spt]
|
||||
aktive.append(3 * p + comp2off[c])
|
||||
|
||||
# Duplikate entfernen
|
||||
out, seen = [], set()
|
||||
for i in aktive:
|
||||
if i not in seen:
|
||||
seen.add(i)
|
||||
out.append(i)
|
||||
return out
|
||||
|
||||
@staticmethod
|
||||
def auswahlmatrix_E(u: int, aktive_unbekannte_indices: Iterable[int]) -> sp.Matrix:
|
||||
E = sp.zeros(u, u)
|
||||
for idx in aktive_unbekannte_indices:
|
||||
i = int(idx)
|
||||
if not (0 <= i < u):
|
||||
raise IndexError(f"Aktiver Index {i} außerhalb [0,{u-1}]")
|
||||
E[i, i] = 1
|
||||
return E
|
||||
|
||||
@staticmethod
|
||||
def raenderungsmatrix_G(
|
||||
x0: sp.Matrix,
|
||||
liste_punktnummern: Sequence[str],
|
||||
*,
|
||||
mit_massstab: bool = True,
|
||||
layout: str = "XYZ",
|
||||
) -> sp.Matrix:
|
||||
if x0.cols != 1:
|
||||
raise ValueError("x0 muss Spaltenvektor sein.")
|
||||
layout = layout.upper()
|
||||
if layout != "XYZ":
|
||||
raise ValueError("Nur layout='XYZ' unterstützt (wie bei euch).")
|
||||
|
||||
u = x0.rows
|
||||
d = 7 if mit_massstab else 6
|
||||
G = sp.zeros(u, d)
|
||||
nP = len(liste_punktnummern)
|
||||
u = x0.rows
|
||||
d = 7 if mit_massstab else 6
|
||||
G = sp.zeros(u, d)
|
||||
|
||||
# --- Translationen ---
|
||||
for i in idx_X:
|
||||
G[i, 0] = 1
|
||||
for i in idx_Y:
|
||||
G[i, 1] = 1
|
||||
for i in idx_Z:
|
||||
G[i, 2] = 1
|
||||
for p in range(nP):
|
||||
ix, iy, iz = 3*p, 3*p+1, 3*p+2
|
||||
xi, yi, zi = x0[ix, 0], x0[iy, 0], x0[iz, 0]
|
||||
|
||||
# --- Rotationen ---
|
||||
# Rotation um X-Achse
|
||||
for iy, iz in zip(idx_Y, idx_Z):
|
||||
zi = x0[iz, 0]
|
||||
yi = x0[iy, 0]
|
||||
G[iy, 3] = -zi
|
||||
G[iz, 3] = yi
|
||||
# Translationen
|
||||
G[ix, 0] = 1
|
||||
G[iy, 1] = 1
|
||||
G[iz, 2] = 1
|
||||
|
||||
# Rotation um Y-Achse
|
||||
for ix, iz in zip(idx_X, idx_Z):
|
||||
zi = x0[iz, 0]
|
||||
xi = x0[ix, 0]
|
||||
G[ix, 4] = zi
|
||||
G[iz, 4] = -xi
|
||||
# Rotationen
|
||||
G[iy, 3] = -zi; G[iz, 3] = yi # Rx
|
||||
G[ix, 4] = zi; G[iz, 4] = -xi # Ry
|
||||
G[ix, 5] = -yi; G[iy, 5] = xi # Rz
|
||||
|
||||
# Rotation um Z-Achse
|
||||
for ix, iy in zip(idx_X, idx_Y):
|
||||
yi = x0[iy, 0]
|
||||
xi = x0[ix, 0]
|
||||
G[ix, 5] = -yi
|
||||
G[iy, 5] = xi
|
||||
# Maßstab
|
||||
if mit_massstab:
|
||||
G[ix, 6] = xi
|
||||
G[iy, 6] = yi
|
||||
G[iz, 6] = zi
|
||||
|
||||
# --- Maßstab ---
|
||||
if mit_massstab:
|
||||
for ix, iy, iz in zip(idx_X, idx_Y, idx_Z):
|
||||
xi = x0[ix, 0]
|
||||
yi = x0[iy, 0]
|
||||
zi = x0[iz, 0]
|
||||
G[ix, 6] = xi
|
||||
G[iy, 6] = yi
|
||||
G[iz, 6] = zi
|
||||
return G
|
||||
return G
|
||||
|
||||
@staticmethod
|
||||
def berechne_dx_geraendert(N: sp.Matrix, n: sp.Matrix, Gi: sp.Matrix) -> sp.Matrix:
|
||||
if N.rows != N.cols:
|
||||
raise ValueError("N muss quadratisch sein.")
|
||||
if n.cols != 1:
|
||||
raise ValueError("n muss Spaltenvektor sein.")
|
||||
if Gi.rows != N.rows:
|
||||
raise ValueError("Gi hat falsche Zeilenzahl.")
|
||||
|
||||
def auswahlmatrix_E(u: int, aktive_unbekannte_indices: Iterable[int]) -> sp.Matrix:
|
||||
E = sp.zeros(u, u)
|
||||
for idx in aktive_unbekannte_indices:
|
||||
E[int(idx), int(idx)] = 1
|
||||
return E
|
||||
u = N.rows
|
||||
d = Gi.cols
|
||||
K = N.row_join(Gi)
|
||||
K = K.col_join(Gi.T.row_join(sp.zeros(d, d)))
|
||||
rhs = n.col_join(sp.zeros(d, 1))
|
||||
sol = K.LUsolve(rhs)
|
||||
return sol[:u, :]
|
||||
|
||||
@staticmethod
|
||||
def weiches_datum(
|
||||
A: sp.Matrix,
|
||||
dl: sp.Matrix,
|
||||
Q_ll: sp.Matrix,
|
||||
x0: sp.Matrix,
|
||||
anschluss_indices: Sequence[int],
|
||||
anschluss_werte: sp.Matrix,
|
||||
Sigma_AA: Optional[sp.Matrix] = None,
|
||||
) -> Tuple[sp.Matrix, sp.Matrix, sp.Matrix]:
|
||||
if dl.cols != 1 or x0.cols != 1:
|
||||
raise ValueError("dl und x0 müssen Spaltenvektoren sein.")
|
||||
if A.rows != dl.rows:
|
||||
raise ValueError("A.rows muss dl.rows entsprechen.")
|
||||
if A.cols != x0.rows:
|
||||
raise ValueError("A.cols muss x0.rows entsprechen.")
|
||||
if Q_ll.rows != Q_ll.cols or Q_ll.rows != A.rows:
|
||||
raise ValueError("Q_ll muss (n×n) sein und zu A.rows passen.")
|
||||
|
||||
def teilspurminimierung_Gi(G: sp.Matrix, E: sp.Matrix) -> sp.Matrix:
|
||||
Gi = E * G
|
||||
return Gi
|
||||
u = A.cols
|
||||
idx = [int(i) for i in anschluss_indices]
|
||||
m = len(idx)
|
||||
|
||||
if anschluss_werte.cols != 1 or anschluss_werte.rows != m:
|
||||
raise ValueError("anschluss_werte muss (m×1) sein.")
|
||||
if Sigma_AA is None:
|
||||
Sigma_AA = sp.eye(m)
|
||||
if Sigma_AA.rows != m or Sigma_AA.cols != m:
|
||||
raise ValueError("Sigma_AA muss (m×m) sein.")
|
||||
|
||||
def berechne_dx_geraendert(N: sp.Matrix, n: sp.Matrix, Gi: sp.Matrix) -> sp.Matrix:
|
||||
u = N.rows
|
||||
d = Gi.shape[1]
|
||||
A_A = sp.zeros(m, u)
|
||||
for r, j in enumerate(idx):
|
||||
if not (0 <= j < u):
|
||||
raise IndexError(f"Anschluss-Index {j} außerhalb [0,{u-1}]")
|
||||
A_A[r, j] = 1
|
||||
|
||||
K = N.row_join(Gi)
|
||||
K = K.col_join(Gi.T.row_join(sp.zeros(d, d)))
|
||||
x0_A = sp.Matrix([[x0[j, 0]] for j in idx])
|
||||
dl_A = anschluss_werte - x0_A
|
||||
|
||||
rhs = n.col_join(sp.zeros(d, 1))
|
||||
A_ext = A.col_join(A_A)
|
||||
dl_ext = dl.col_join(dl_A)
|
||||
|
||||
sol = K.LUsolve(rhs)
|
||||
dx = sol[:u, :]
|
||||
return dx
|
||||
Q_ext = sp.zeros(Q_ll.rows + m, Q_ll.cols + m)
|
||||
Q_ext[:Q_ll.rows, :Q_ll.cols] = Q_ll
|
||||
Q_ext[Q_ll.rows:, Q_ll.cols:] = Sigma_AA
|
||||
|
||||
return A_ext, dl_ext, Q_ext
|
||||
|
||||
@@ -1,29 +1,62 @@
|
||||
from Datumsfestlegung import *
|
||||
from Stochastisches_Modell import StochastischesModell
|
||||
from Netzqualität_Genauigkeit import Genauigkeitsmaße
|
||||
from Datumsfestlegung import Datumsfestlegung
|
||||
import sympy as sp
|
||||
import Export
|
||||
import Netzqualität_Genauigkeit
|
||||
|
||||
|
||||
def ausgleichung_global(A, dl, stoch_modell: StochastischesModell):
|
||||
def ausgleichung_global(
|
||||
A: sp.Matrix,
|
||||
dl: sp.Matrix,
|
||||
Q_ll: sp.Matrix,
|
||||
x0: sp.Matrix,
|
||||
idx_X, idx_Y, idx_Z,
|
||||
anschluss_indices,
|
||||
anschluss_werte,
|
||||
Sigma_AA,
|
||||
):
|
||||
# 1) Datumsfestlegung (weiches Datum) System erweitern
|
||||
A_ext, dl_ext, Q_ext = Datumsfestlegung.weiches_datum(
|
||||
A=A,
|
||||
dl=dl,
|
||||
Q_ll=Q_ll,
|
||||
x0=x0,
|
||||
anschluss_indices=anschluss_indices,
|
||||
anschluss_werte=anschluss_werte,
|
||||
Sigma_AA=Sigma_AA,
|
||||
)
|
||||
|
||||
#Q_ll, P = stoch_modell.berechne_Qll_P() #Kofaktormatrix und P-Matrix
|
||||
P = sp.eye(A.shape[0])
|
||||
N = A.T * P * A #Normalgleichungsmatrix N
|
||||
Q_xx = N.inv() #Kofaktormatrix der Unbekannten Qxx
|
||||
n = A.T * P * dl #Absolutgliedvektor n
|
||||
# 2) Gewichtsmatrix P
|
||||
P = StochastischesModell.berechne_P(Q_ext)
|
||||
|
||||
dx = N.LUsolve(n) #Zuschlagsvektor dx
|
||||
# 3) Normalgleichungsmatrix N und Absolutgliedvektor n
|
||||
N = A_ext.T * P * A_ext
|
||||
n = A_ext.T * P * dl_ext
|
||||
|
||||
v = dl - A * dx #Residuenvektor v
|
||||
# 4) Zuschlagsvektor dx
|
||||
dx = N.LUsolve(n)
|
||||
|
||||
# 5) Residuenvektor v
|
||||
v = dl - A * dx
|
||||
|
||||
# 6) Kofaktormatrix der Unbekannten Q_xx
|
||||
Q_xx = StochastischesModell.berechne_Q_xx(N)
|
||||
|
||||
# 7) Kofaktormatrix der Beobachtungen Q_ll_dach
|
||||
Q_ll_dach = A * Q_xx * A.T
|
||||
Q_vv = stoch_modell.berechne_Qvv(A, P, Q_xx) #Kofaktormatrix der Verbesserungen Qvv
|
||||
R = stoch_modell.berechne_R(Q_vv, P) #Redundanzmatrix R
|
||||
r = stoch_modell.berechne_r(R) #Redundanzanteile als Vektor r
|
||||
redundanzanteile = A.shape[0] - A.shape[1] #n-u+d
|
||||
soaposteriori = Netzqualität_Genauigkeit.Genauigkeitsmaße.s0apost(v, P, redundanzanteile)
|
||||
|
||||
# 8) Kofaktormatrix der Verbesserungen Q_vv
|
||||
Q_vv = StochastischesModell.berechne_Qvv(A, P, Q_xx)
|
||||
|
||||
# 9) Redundanzmatrix R und Redundanzanteile r
|
||||
R = StochastischesModell.berechne_R(Q_vv, P) #Redundanzmatrix R
|
||||
r = StochastischesModell.berechne_r(R) #Redundanzanteile als Vektor r
|
||||
redundanzanteile = A.shape[0] - A.shape[1] #n-u+d
|
||||
|
||||
# 10) s0 a posteriori
|
||||
soaposteriori = Genauigkeitsmaße.s0apost(v, P, redundanzanteile)
|
||||
|
||||
# 11) Ausgabe
|
||||
dict_ausgleichung = {
|
||||
"dx": dx,
|
||||
"v": v,
|
||||
@@ -38,68 +71,65 @@ def ausgleichung_global(A, dl, stoch_modell: StochastischesModell):
|
||||
}
|
||||
|
||||
Export.Export.ausgleichung_to_datei(r"Zwischenergebnisse\Ausgleichung_Iteration0.csv", dict_ausgleichung)
|
||||
|
||||
return dict_ausgleichung, dx
|
||||
|
||||
|
||||
|
||||
|
||||
def ausgleichung_lokal(
|
||||
A: sp.Matrix,
|
||||
dl: sp.Matrix,
|
||||
stoch_modell: StochastischesModell,
|
||||
Q_ll: sp.Matrix,
|
||||
x0: sp.Matrix,
|
||||
idx_X, idx_Y, idx_Z,
|
||||
aktive_unbekannte_indices,
|
||||
mit_massstab: bool = True,
|
||||
):
|
||||
# 1) Gewichte
|
||||
Q_ll, P = stoch_modell.berechne_Qll_P()
|
||||
# Debug-Option:
|
||||
# P = sp.eye(A.rows)
|
||||
# 1) Gewichtsmatrix P
|
||||
P = StochastischesModell.berechne_P(Q_ll)
|
||||
|
||||
# 2) Normalgleichungen
|
||||
# 2) Normalgleichungsmatrix N und Absolutgliedvektor n
|
||||
N = A.T * P * A
|
||||
n = A.T * P * dl
|
||||
|
||||
# 3) Datum (G, E, Gi)
|
||||
G = raenderungsmatrix_G(x0, idx_X, idx_Y, idx_Z, mit_massstab=mit_massstab)
|
||||
E = auswahlmatrix_E(u=A.cols, aktive_unbekannte_indices=aktive_unbekannte_indices)
|
||||
# 3) Datumsfestlegung (Teilspurminimierung)
|
||||
G = Datumsfestlegung.raenderungsmatrix_G(x0, liste_punktnummern, mit_massstab=mit_massstab)
|
||||
aktive = Datumsfestlegung.datumskomponenten(auswahl, liste_punktnummern)
|
||||
E = Datumsfestlegung.auswahlmatrix_E(u=A.cols, aktive_unbekannte_indices=aktive)
|
||||
Gi = E * G
|
||||
|
||||
# 4) Geränderte Lösung (dx)
|
||||
dx = berechne_dx_geraendert(N, n, Gi)
|
||||
# 4) Zuschlagsvektor dx
|
||||
dx = Datumsfestlegung.berechne_dx_geraendert(N, n, Gi)
|
||||
|
||||
# 5) Residuen
|
||||
# 5) Residuenvektor v
|
||||
v = dl - A * dx
|
||||
|
||||
# 6) KORREKTE Q_xx für gerändertes Problem:
|
||||
# Q_xx = N^{-1} - N^{-1}Gi (Gi^T N^{-1} Gi)^{-1} Gi^T N^{-1}
|
||||
# numerisch besser via LUsolve statt inv:
|
||||
N_inv = N.inv() # wenn N groß ist, kann man das unten auch ohne inv machen (siehe Hinweis)
|
||||
# 6) Kofaktormatrix der Unbekannten Q_xx
|
||||
N_inv = N.inv()
|
||||
N_inv_G = N_inv * Gi
|
||||
S = Gi.T * N_inv_G
|
||||
S_inv = S.inv()
|
||||
Q_xx = N_inv - N_inv_G * S_inv * N_inv_G.T
|
||||
|
||||
# 7) Q_lhat_lhat und Q_vv
|
||||
# 7) Kofaktormatrix der Beobachtungen Q_ll_dach
|
||||
Q_lhat_lhat = A * Q_xx * A.T
|
||||
|
||||
# 8) Kofaktormatrix der Verbesserungen Q_vv
|
||||
Q_vv = P.inv() - Q_lhat_lhat
|
||||
|
||||
# 8) Redundanzmatrix und -anteile
|
||||
# 9) Redundanzmatrix R, Redundanzanteile r, Redundanz
|
||||
R = Q_vv * P
|
||||
r_vec = sp.Matrix(R.diagonal())
|
||||
|
||||
# 9) Freiheitsgrade (Redundanz gesamt)
|
||||
n_beob = A.rows
|
||||
u = A.cols
|
||||
d = Gi.shape[1]
|
||||
r_gesamt = n_beob - u + d
|
||||
|
||||
# 10) sigma0 a posteriori
|
||||
omega = float((v.T * P * v)[0, 0])
|
||||
sigma0_hat = (omega / float(r_gesamt)) ** 0.5
|
||||
# 10) s0 a posteriori
|
||||
sigma0_apost = Genauigkeitsmaße.s0apost(v, P, r_gesamt)
|
||||
|
||||
return {
|
||||
# 11) Ausgabe
|
||||
dict_ausgleichung_lokal = {
|
||||
"dx": dx,
|
||||
"v": v,
|
||||
"Q_ll": Q_ll,
|
||||
@@ -111,7 +141,10 @@ def ausgleichung_lokal(
|
||||
"R": R,
|
||||
"r": r_vec,
|
||||
"r_gesamt": r_gesamt,
|
||||
"sigma0_hat": sigma0_hat,
|
||||
"sigma0_apost": sigma0_apost,
|
||||
"G": G,
|
||||
"Gi": Gi,
|
||||
}
|
||||
}
|
||||
|
||||
Export.Export.ausgleichung_to_datei(r"Zwischenergebnisse\Ausgleichung_Iteration0_lokal.csv", dict_ausgleichung_lokal)
|
||||
return dict_ausgleichung_lokal, dx
|
||||
@@ -49,11 +49,18 @@ class StochastischesModell:
|
||||
return Q_ll
|
||||
|
||||
|
||||
def berechne_P(Q_ll):
|
||||
def berechne_P(Q_ll: sp.Matrix) -> sp.Matrix:
|
||||
P = Q_ll.inv()
|
||||
return P
|
||||
|
||||
|
||||
def berechne_Q_xx(N: sp.Matrix) -> sp.Matrix:
|
||||
if N.rows != N.cols:
|
||||
raise ValueError("N muss eine quadratische Matrix sein")
|
||||
Q_xx = N.inv()
|
||||
return Q_xx
|
||||
|
||||
|
||||
def berechne_Qvv(self, A: sp.Matrix, P: sp.Matrix, Q_xx: sp.Matrix) -> sp.Matrix:
|
||||
Q_vv = P.inv() - A * Q_xx * A.T
|
||||
return Q_vv #Kofaktormatrix der Beobachtungsresiduen
|
||||
|
||||
Reference in New Issue
Block a user