Spaces:
Sleeping
Sleeping
Update src/visualization.py
#32
by axelrvite - opened
- src/visualization.py +83 -22
src/visualization.py
CHANGED
|
@@ -1,4 +1,4 @@
|
|
| 1 |
-
#!/usr/bin/env python3
|
| 2 |
# -*- coding: utf-8 -*-
|
| 3 |
"""
|
| 4 |
Module de visualisation pour le projet d'IA solaire.
|
|
@@ -17,7 +17,9 @@ from matplotlib.colors import LinearSegmentedColormap
|
|
| 17 |
import matplotlib.cm as cm
|
| 18 |
from mpl_toolkits.mplot3d import Axes3D
|
| 19 |
from pvlib.pvsystem import i_from_v
|
| 20 |
-
import
|
|
|
|
|
|
|
| 21 |
|
| 22 |
|
| 23 |
def iv_curve(voltage, rsh, rs, nNsVth, J0, Jph):
|
|
@@ -38,17 +40,13 @@ class SolarCellVisualizer:
|
|
| 38 |
pour la prédiction des paramètres des cellules solaires.
|
| 39 |
"""
|
| 40 |
|
| 41 |
-
def __init__(self, results_dir=
|
| 42 |
"""
|
| 43 |
Initialise le visualiseur.
|
| 44 |
|
| 45 |
Args:
|
| 46 |
results_dir (str): Répertoire pour sauvegarder les résultats.
|
| 47 |
"""
|
| 48 |
-
if results_dir is None:
|
| 49 |
-
results_dir = os.path.join(tempfile.gettempdir(), "results") # ✅ Dossier temporaire
|
| 50 |
-
|
| 51 |
-
os.makedirs(results_dir, exist_ok=True)
|
| 52 |
self.results_dir = results_dir
|
| 53 |
|
| 54 |
# Constants
|
|
@@ -90,9 +88,11 @@ class SolarCellVisualizer:
|
|
| 90 |
ax1.legend(fontsize=12)
|
| 91 |
|
| 92 |
# Tracer l'erreur absolue moyenne
|
|
|
|
|
|
|
| 93 |
ax2 = plt.subplot(gs[0, 1])
|
| 94 |
-
ax2.plot(
|
| 95 |
-
ax2.plot(
|
| 96 |
ax2.set_title('Évolution de l\'erreur absolue moyenne', fontsize=14, fontweight='bold')
|
| 97 |
ax2.set_xlabel('Époque', fontsize=12)
|
| 98 |
ax2.set_ylabel('MAE', fontsize=12)
|
|
@@ -116,7 +116,7 @@ class SolarCellVisualizer:
|
|
| 116 |
# Calculer les statistiques
|
| 117 |
best_epoch = np.argmin(history['val_loss'])
|
| 118 |
best_val_loss = history['val_loss'][best_epoch]
|
| 119 |
-
best_val_mae =
|
| 120 |
|
| 121 |
# Créer le tableau
|
| 122 |
table_data = [
|
|
@@ -571,9 +571,8 @@ class SolarCellVisualizer:
|
|
| 571 |
else:
|
| 572 |
plt.close()
|
| 573 |
|
| 574 |
-
def test_all_data(self,data_processor, model, X_test, y_test):
|
| 575 |
param_names = ['Rsh', 'Rs', 'n', 'J0', 'Jph']
|
| 576 |
-
|
| 577 |
y_pred = model.predict(X_test)
|
| 578 |
|
| 579 |
# Inverser la normalisation
|
|
@@ -591,9 +590,19 @@ class SolarCellVisualizer:
|
|
| 591 |
params_pred = y_pred_orig[i]
|
| 592 |
params = {}
|
| 593 |
for i, param in enumerate(param_names):
|
| 594 |
-
|
|
|
|
|
|
|
| 595 |
I_model = -iv_curve(v_values, params['Rsh'], params['Rs'], params['n']*self.Vth, params['J0'], params['Jph'])
|
| 596 |
-
I_reel =
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 597 |
|
| 598 |
ssd = np.mean(np.square(I_model - I_reel), axis=0)
|
| 599 |
if not np.isnan(ssd):
|
|
@@ -603,8 +612,8 @@ class SolarCellVisualizer:
|
|
| 603 |
|
| 604 |
for i, param in enumerate(param_names):
|
| 605 |
real_val = params_real[i]
|
| 606 |
-
pred_val =
|
| 607 |
-
logger.info(f"vraies infos {real_val} {pred_val}")
|
| 608 |
|
| 609 |
rel_error[param] += np.abs((real_val - pred_val) / (real_val)) * 100
|
| 610 |
for i, param in enumerate(param_names):
|
|
@@ -613,23 +622,74 @@ class SolarCellVisualizer:
|
|
| 613 |
ssd_mean = ssd_mean / nb_good_samples
|
| 614 |
logger.info(f"SSD moyen pour les échantillons de test: {ssd_mean:.4g}")
|
| 615 |
return rel_error, ssd_mean
|
|
|
|
| 616 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 617 |
|
| 618 |
-
def plot_comparaison_courbes(self,data_processor, model, X_test, lst_I_real, lst_V, Plot =True):
|
| 619 |
|
|
|
|
|
|
|
| 620 |
param_names = ['Rsh', 'Rs', 'n', 'J0', 'Jph']
|
| 621 |
-
|
| 622 |
y_pred = model.predict(X_test)
|
| 623 |
# Inverser la normalisation
|
| 624 |
y_pred_orig = data_processor.inverse_transform_params(y_pred)
|
|
|
|
| 625 |
for i in range (len(X_test)):
|
| 626 |
params = {'Rsh':0, 'Rs':0, 'n':0, 'J0':0, 'Jph':0}
|
| 627 |
params_pred = y_pred_orig[i]
|
| 628 |
for j, param in enumerate(param_names):
|
| 629 |
params[param] = params_pred[j]
|
| 630 |
-
|
| 631 |
I_model = -iv_curve(lst_V[i], params['Rsh'], params['Rs'], params['n']*self.Vth, params['J0'], params['Jph'])
|
| 632 |
-
|
| 633 |
if Plot:
|
| 634 |
plt.plot(lst_V[i], I_model, label='Courbe I-V prédite', color='red')
|
| 635 |
plt.plot(lst_V[i], lst_I_real[i], label='Courbe I-V réelle', color='blue')
|
|
@@ -640,6 +700,7 @@ class SolarCellVisualizer:
|
|
| 640 |
plt.legend(fontsize=12)
|
| 641 |
plt.show()
|
| 642 |
return params, I_model
|
|
|
|
| 643 |
|
| 644 |
def comparaison_courbes(self,data_processor, model, X_test, lst_I_real, lst_V):
|
| 645 |
|
|
@@ -651,8 +712,8 @@ class SolarCellVisualizer:
|
|
| 651 |
for i in range (len(X_test)):
|
| 652 |
params = {'Rsh':0, 'Rs':0, 'n':0, 'J0':0, 'Jph':0}
|
| 653 |
params_pred = y_pred_orig[i]
|
| 654 |
-
for
|
| 655 |
-
params[param] = params_pred[
|
| 656 |
|
| 657 |
I_model = -iv_curve(lst_V, params['Rsh'], params['Rs'], params['n']*self.Vth, params['J0'], params['Jph'])
|
| 658 |
return params, I_model
|
|
|
|
| 1 |
+
#!/usr/bin/env python3
|
| 2 |
# -*- coding: utf-8 -*-
|
| 3 |
"""
|
| 4 |
Module de visualisation pour le projet d'IA solaire.
|
|
|
|
| 17 |
import matplotlib.cm as cm
|
| 18 |
from mpl_toolkits.mplot3d import Axes3D
|
| 19 |
from pvlib.pvsystem import i_from_v
|
| 20 |
+
from scipy.optimize import curve_fit
|
| 21 |
+
from scipy.interpolate import interp1d
|
| 22 |
+
|
| 23 |
|
| 24 |
|
| 25 |
def iv_curve(voltage, rsh, rs, nNsVth, J0, Jph):
|
|
|
|
| 40 |
pour la prédiction des paramètres des cellules solaires.
|
| 41 |
"""
|
| 42 |
|
| 43 |
+
def __init__(self, results_dir='../results'):
|
| 44 |
"""
|
| 45 |
Initialise le visualiseur.
|
| 46 |
|
| 47 |
Args:
|
| 48 |
results_dir (str): Répertoire pour sauvegarder les résultats.
|
| 49 |
"""
|
|
|
|
|
|
|
|
|
|
|
|
|
| 50 |
self.results_dir = results_dir
|
| 51 |
|
| 52 |
# Constants
|
|
|
|
| 88 |
ax1.legend(fontsize=12)
|
| 89 |
|
| 90 |
# Tracer l'erreur absolue moyenne
|
| 91 |
+
mae_moy = np.mean([history['output_Rsh_mae'],history['output_Rs_mae'],history['output_n_mae'],history['output_J0_mae'],history['output_Jph_mae'] ], axis=0)
|
| 92 |
+
val_mae_moy = np.mean([history['val_output_Rsh_mae'],history['val_output_Rs_mae'],history['val_output_n_mae'],history['val_output_J0_mae'],history['val_output_Jph_mae'] ], axis=0)
|
| 93 |
ax2 = plt.subplot(gs[0, 1])
|
| 94 |
+
ax2.plot(mae_moy, 'g-', linewidth=2, label='Entraînement')
|
| 95 |
+
ax2.plot(val_mae_moy, 'm-', linewidth=2, label='Validation')
|
| 96 |
ax2.set_title('Évolution de l\'erreur absolue moyenne', fontsize=14, fontweight='bold')
|
| 97 |
ax2.set_xlabel('Époque', fontsize=12)
|
| 98 |
ax2.set_ylabel('MAE', fontsize=12)
|
|
|
|
| 116 |
# Calculer les statistiques
|
| 117 |
best_epoch = np.argmin(history['val_loss'])
|
| 118 |
best_val_loss = history['val_loss'][best_epoch]
|
| 119 |
+
best_val_mae = val_mae_moy[best_epoch]
|
| 120 |
|
| 121 |
# Créer le tableau
|
| 122 |
table_data = [
|
|
|
|
| 571 |
else:
|
| 572 |
plt.close()
|
| 573 |
|
| 574 |
+
def test_all_data(self,data_processor, model, X_test, y_test, Fit = False, Plot = False):
|
| 575 |
param_names = ['Rsh', 'Rs', 'n', 'J0', 'Jph']
|
|
|
|
| 576 |
y_pred = model.predict(X_test)
|
| 577 |
|
| 578 |
# Inverser la normalisation
|
|
|
|
| 590 |
params_pred = y_pred_orig[i]
|
| 591 |
params = {}
|
| 592 |
for i, param in enumerate(param_names):
|
| 593 |
+
params[param] = params_pred[i]
|
| 594 |
+
if Fit:
|
| 595 |
+
params, I_after_fit, ssd = self.fit_after_model(X_orig[:, 1], X_orig[:, 0], params)
|
| 596 |
I_model = -iv_curve(v_values, params['Rsh'], params['Rs'], params['n']*self.Vth, params['J0'], params['Jph'])
|
| 597 |
+
I_reel = X_orig[:, 1]
|
| 598 |
+
if Plot:
|
| 599 |
+
plt.plot(v_values, I_model, label='Modèle')
|
| 600 |
+
plt.plot(v_values, I_reel, label='Réel')
|
| 601 |
+
plt.title(f'Comparaison des courbes I-V - Échantillon {i}')
|
| 602 |
+
plt.xlabel('Tension (V)')
|
| 603 |
+
plt.ylabel('Courant (I)')
|
| 604 |
+
plt.legend()
|
| 605 |
+
plt.show()
|
| 606 |
|
| 607 |
ssd = np.mean(np.square(I_model - I_reel), axis=0)
|
| 608 |
if not np.isnan(ssd):
|
|
|
|
| 612 |
|
| 613 |
for i, param in enumerate(param_names):
|
| 614 |
real_val = params_real[i]
|
| 615 |
+
pred_val = params[param]
|
| 616 |
+
#logger.info(f"vraies infos {real_val} {pred_val}")
|
| 617 |
|
| 618 |
rel_error[param] += np.abs((real_val - pred_val) / (real_val)) * 100
|
| 619 |
for i, param in enumerate(param_names):
|
|
|
|
| 622 |
ssd_mean = ssd_mean / nb_good_samples
|
| 623 |
logger.info(f"SSD moyen pour les échantillons de test: {ssd_mean:.4g}")
|
| 624 |
return rel_error, ssd_mean
|
| 625 |
+
|
| 626 |
|
| 627 |
+
def fit_after_model(self, I, V, params):
|
| 628 |
+
k = 1.3806503e-23 # Boltzmann constant (J/K)
|
| 629 |
+
q = 1.60217646e-19 # Elementary charge (C)
|
| 630 |
+
|
| 631 |
+
#self.Tmes = int(chemin.split("\\")[-1].split("_")[1]) + 273.15
|
| 632 |
+
Tmes = 300
|
| 633 |
+
|
| 634 |
+
Vth = k * Tmes / q
|
| 635 |
+
|
| 636 |
+
params_to_fit = { "Rsh": params['Rsh'],
|
| 637 |
+
"Rs": params['Rs'],
|
| 638 |
+
"nVth": params['n']*Vth,
|
| 639 |
+
"J0": params['J0'],
|
| 640 |
+
"Jph": params['Jph']}
|
| 641 |
+
bounds = {
|
| 642 |
+
"Rsh": [1, 4000],
|
| 643 |
+
"Rs": [0.1, 5],
|
| 644 |
+
"nVth": [0.8*Vth, 3*Vth],
|
| 645 |
+
"J0": [1e-20, 1e-5],
|
| 646 |
+
}
|
| 647 |
+
params_to_fit["Jph"] = -I[0]
|
| 648 |
+
|
| 649 |
+
if params_to_fit["Jph"] > 0:
|
| 650 |
+
low_bounds = [bounds["Rsh"][0],bounds["Rs"][0], bounds["nVth"][0], bounds["J0"][0], params_to_fit["Jph"] * 0.9]
|
| 651 |
+
high_bounds = [bounds["Rsh"][1],bounds["Rs"][1], bounds["nVth"][1], bounds["J0"][1],params_to_fit["Jph"] * 1.1]
|
| 652 |
+
else:
|
| 653 |
+
low_bounds = [bounds["Rsh"][0],bounds["Rs"][0], bounds["nVth"][0], bounds["J0"][0], params_to_fit["Jph"] * 1.1]
|
| 654 |
+
high_bounds = [bounds["Rsh"][1],bounds["Rs"][1], bounds["nVth"][1], bounds["J0"][1],params_to_fit["Jph"] * 0.9]
|
| 655 |
+
#print("low_bounds", low_bounds)
|
| 656 |
+
#print("high_bounds", high_bounds)
|
| 657 |
+
|
| 658 |
+
# Ajustement automatique des paramètres s'ils dépassent les bornes
|
| 659 |
+
params_to_fit["Rsh"] = min(max(params_to_fit["Rsh"], bounds["Rsh"][0]), bounds["Rsh"][1])
|
| 660 |
+
params_to_fit["Rs"] = min(max(params_to_fit["Rs"], bounds["Rs"][0]), bounds["Rs"][1])
|
| 661 |
+
params_to_fit["nVth"] = min(max(params_to_fit["nVth"], bounds["nVth"][0]), bounds["nVth"][1])
|
| 662 |
+
params_to_fit["J0"] = min(max(params_to_fit["J0"], bounds["J0"][0]), bounds["J0"][1])
|
| 663 |
+
|
| 664 |
+
popt, pcov = curve_fit(iv_curve, V, -I,
|
| 665 |
+
p0=list(params_to_fit.values()), bounds=(low_bounds, high_bounds), maxfev=10000)
|
| 666 |
+
param_fit = { "Rsh": popt[0],
|
| 667 |
+
"Rs": popt[1],
|
| 668 |
+
"nVth": popt[2],
|
| 669 |
+
"J0": popt[3],
|
| 670 |
+
"Jph":popt[4]}
|
| 671 |
+
param_fit["n"] = param_fit["nVth"] / Vth
|
| 672 |
+
I_after_fit = -iv_curve(V, param_fit['Rsh'], param_fit['Rs'], param_fit['n']*Vth, param_fit['J0'], param_fit['Jph'])
|
| 673 |
+
ssd = np.mean((I - I_after_fit) ** 2)
|
| 674 |
+
print("SSD après ajustement:", ssd)
|
| 675 |
+
return param_fit,I_after_fit, ssd
|
| 676 |
|
|
|
|
| 677 |
|
| 678 |
+
def plot_comparaison_courbes(self,data_processor, model, X_test, lst_I_real, lst_V, Plot =True):
|
| 679 |
+
|
| 680 |
param_names = ['Rsh', 'Rs', 'n', 'J0', 'Jph']
|
| 681 |
+
|
| 682 |
y_pred = model.predict(X_test)
|
| 683 |
# Inverser la normalisation
|
| 684 |
y_pred_orig = data_processor.inverse_transform_params(y_pred)
|
| 685 |
+
|
| 686 |
for i in range (len(X_test)):
|
| 687 |
params = {'Rsh':0, 'Rs':0, 'n':0, 'J0':0, 'Jph':0}
|
| 688 |
params_pred = y_pred_orig[i]
|
| 689 |
for j, param in enumerate(param_names):
|
| 690 |
params[param] = params_pred[j]
|
|
|
|
| 691 |
I_model = -iv_curve(lst_V[i], params['Rsh'], params['Rs'], params['n']*self.Vth, params['J0'], params['Jph'])
|
| 692 |
+
|
| 693 |
if Plot:
|
| 694 |
plt.plot(lst_V[i], I_model, label='Courbe I-V prédite', color='red')
|
| 695 |
plt.plot(lst_V[i], lst_I_real[i], label='Courbe I-V réelle', color='blue')
|
|
|
|
| 700 |
plt.legend(fontsize=12)
|
| 701 |
plt.show()
|
| 702 |
return params, I_model
|
| 703 |
+
|
| 704 |
|
| 705 |
def comparaison_courbes(self,data_processor, model, X_test, lst_I_real, lst_V):
|
| 706 |
|
|
|
|
| 712 |
for i in range (len(X_test)):
|
| 713 |
params = {'Rsh':0, 'Rs':0, 'n':0, 'J0':0, 'Jph':0}
|
| 714 |
params_pred = y_pred_orig[i]
|
| 715 |
+
for j, param in enumerate(param_names):
|
| 716 |
+
params[param] = params_pred[j]
|
| 717 |
|
| 718 |
I_model = -iv_curve(lst_V, params['Rsh'], params['Rs'], params['n']*self.Vth, params['J0'], params['Jph'])
|
| 719 |
return params, I_model
|