forked from benjamin2044/PV_fault_Python
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathPNN_Solar_model.py
222 lines (181 loc) · 9.26 KB
/
PNN_Solar_model.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
### 3 Faults selection ###
## Solar_data_Experiment
########### Part1 (DATA PRE-PROCESSING) ###############
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sb
dataset = pd.read_csv('Solar_categorical.csv')
X = dataset.iloc[:3000, 0:7].values
y = dataset.iloc[:3000, 7].values
#print(y)
"""
###### Input data Data Visualization #####
plt.xlabel('Sample size', fontsize = 10)
plt.ylabel('Current (A)', fontsize = 10)
plt.title('Summer features', fontsize = 12)
plt.plot(X[:250-1, 0], label='Normal Sunny') # Summer normal sunny
plt.plot(X[251:501-1, 0], label='Normal Cloudy') # Summer normal cloudy
plt.plot(X[1002:1249-1, 1], label='Line-line Sunny') # Summer line-line cloudy
plt.plot(X[1251:1501-1, 0], label='Line-line Cloudy') # Summer line-line cloudy
plt.legend(fancybox=True, shadow=True)
plt.xlabel('Sample size', fontsize = 10)
plt.ylabel('Current (A)', fontsize = 10)
plt.title('Winter features', fontsize = 12)
plt.plot(X[1502:1751-1, 0], label='Normal Sunny') # Winter normal sunny
plt.plot(X[1752:2001, 0], label='Normal Cloudy') # Winter normal cloudy
plt.plot(X[2502:2750, 1], label='Line-line Sunny') # Winter line-line sunny
plt.plot(X[2751:3001, 1], label='Line-line Cloudy') # Winter normal cloudy (temp 4 deg to -5,)
plt.legend(fancybox=True, shadow=True)
#Plotting in subplots
fig, (ax1, ax2) = plt.subplots(1,2)
fig.suptitle( "Summer and winter features")
ax1.plot(X[:250-1, 0], label='Normal Sunny')
ax1.plot(X[251:501-1, 0], label='Normal Cloudy') # Summer normal cloudy
ax1.plot(X[1002:1249-1, 1], label='Line-line Sunny') # Summer line-line cloudy
ax1.plot(X[1251:1501-1, 0], label='Line-line Cloudy') # Summer line-line cloudy
ax1.legend()
ax2.plot(X[1502:1751-1, 0], label='Normal Sunny') # Winter normal sunny
ax2.plot(X[1752:2001, 0], label='Normal Cloudy') # Winter normal cloudy
ax2.plot(X[2502:2750, 1], label='Line-line Sunny') # Winter line-line sunny
ax2.plot(X[2751:3001, 1], label='Line-line Cloudy') # Winter normal cloudy (temp 4 deg to -5,)
ax2.legend()
"""
########## Label Encoding categorical data ###########
from sklearn.preprocessing import LabelEncoder
from keras.utils import to_categorical
encoder= LabelEncoder()
X[:,6] = encoder.fit_transform(X[:, 6])
y = encoder.fit_transform(y)
#y_original = encoder.inverse_transform(y_encoded)
y = to_categorical(y)
##########################################################################
# Splitting the dataset into the Training set and Test set
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state=0)
#####################################################################################
# Feature Scaling (To scale all variables to similar scale)
from sklearn.preprocessing import StandardScaler
sc = StandardScaler()
X_train = sc.fit_transform(X_train)
X_test = sc.transform(X_test)
############################################################################
########### Part2 (DEEP NEURAL NETWORK)/Multiple Layer Perceptron(MLP)#########################
from keras.models import Sequential ##module to initialize ANN
from keras.layers import Dense #module required to build layers
DNN_model = Sequential()
DNN_model.add(Dense(input_dim = 7, units = 8, kernel_initializer = 'uniform', activation = 'relu'))
DNN_model.add(Dense(units = 8, kernel_initializer = 'uniform', activation = 'relu'))
DNN_model.add(Dense(units = 3, kernel_initializer = 'uniform', activation = 'softmax'))
DNN_model.compile(optimizer = 'adam', loss = 'categorical_crossentropy', metrics = ['accuracy'])
print(DNN_model.summary())
history = DNN_model.fit(X_train, y_train, batch_size = 5, epochs = 200,
validation_data=(X_test, y_test), shuffle=True)
print(history.history.keys())
######Visualizing model: train & test accuracy and loss ############
"""
plt.plot(history.history['accuracy'])
plt.plot(history.history['val_accuracy'])
plt.title('model accuracy')
plt.ylabel('accuracy')
plt.xlabel('epoch')
plt.legend(['train', 'test'])
plt.plot(history.history['loss']) # summarize history for loss
plt.plot(history.history['val_loss'])
plt.title('model loss')
plt.ylabel('loss')
plt.xlabel('epoch')
plt.legend(['train', 'test'], loc='upper right')
"""
################### Part3 (Making predictions and evaluating the model)#################
y_pred = DNN_model.predict(X_test) # Predicting the Test set results
y_pred = (y_pred > 0.95)
y_pred_label = encoder.inverse_transform(np.argmax(y_pred, axis=1))
print(y_pred_label)
y_test_label = encoder.inverse_transform(np.argmax(y_test, axis=1))
print(y_test_label)
#Making confusion matrix that checks accuracy of the model
from sklearn.metrics import confusion_matrix
cm = confusion_matrix(y_test_label, y_pred_label)
#### Visualizing Confusion Matrix ########
cm_fig = pd.DataFrame(cm, columns=np.unique(y_test_label), index=np.unique(y_test_label))
sb.set(font_scale=1.3)
sb.heatmap(cm_fig, cmap="RdBu_r", annot=True, annot_kws={"size":15}, fmt='.2f')
plt.xlabel('Predicted label')
plt.ylabel('True label')
plt.title('Confusion Matrix')
#####################################################################
######### Predicting with new data ##########
new_pred1 = DNN_model.predict(sc.transform(np.array([[5, 6.2, 92, 93, 85, 16, 1]])))
new_pred1_original = encoder.inverse_transform([np.argmax(new_pred1)])
print(new_pred1_original)
new_pred2 = DNN_model.predict(sc.transform(np.array([[0, 3.5, 0, 95, 96, 12, 1]])))
new_pred2_original = encoder.inverse_transform([np.argmax(new_pred2)])
print(new_pred2_original)
new_pred3 = DNN_model.predict(sc.transform(np.array([[3.7, 0.2, 85, 65, 78, 7, 0]])))
new_pred3_original = encoder.inverse_transform([np.argmax(new_pred3)])
print(new_pred3_original)
#################################################################################
###Saving the model
DNN_model.save('fault_model.model')
### Opening the saved model
from keras.models import load_model
new_model = load_model('fault_model.model')
new_model.summary()
###### Predictiong with new data
new_mod_test = new_model.predict(sc.transform(np.array([[2.1, 3.1, 90, 83, 88, 15, 1]])))
new_mod_test_original = encoder.inverse_transform([np.argmax(new_mod_test)])
print(new_mod_test_original)
#####################################################
#Evaluating, Improving and Tuning the ANN
# Evaluating the ANN
from keras.wrappers.scikit_learn import KerasClassifier
from sklearn.model_selection import cross_val_score
from keras.models import Sequential
from keras.layers import Dense
# to check bias-variance tradeoff
def build_classifier():
model = Sequential()
model.add(Dense(input_dim = 7, units = 8, kernel_initializer = 'uniform', activation = 'relu'))
model.add(Dense(units = 8, kernel_initializer = 'uniform', activation = 'relu'))
model.add(Dense(units = 3, kernel_initializer = 'uniform', activation = 'softmax'))
model.compile(optimizer = 'adam', loss = 'categorical_crossentropy', metrics = ['accuracy'])
return model
model = KerasClassifier(build_fn = build_classifier, batch_size = 5, epochs = 200)
accuracies = cross_val_score(estimator = model, X = X_train, y = y_train, cv = 5) #CV = K-fold cross val. splits
mean = accuracies.mean()
variance = accuracies.std()
print("Average accuracy:", mean, "Variance:", variance)
# Improving the ANN
# Dropout Regularization to reduce overfitting if needed
from keras.layers import Dropout
model = Sequential()
model.add(Dense(input_dim = 7, units = 8, kernel_initializer = 'uniform', activation = 'relu'))
model.add(Dropout(rate = 0.1))
model.add(Dense(units = 8, kernel_initializer = 'uniform', activation = 'relu'))
model.add(Dropout(rate = 0.2))
model.add(Dense(units = 3, kernel_initializer = 'uniform', activation = 'softmax'))
model.compile(optimizer = 'adam', loss = 'categorical_crossentropy', metrics = ['accuracy'])
model.fit(X_train, y_train, batch_size = 5, epochs = 200)
y_pred_drop = model.predict(X_test)
y_pred_drop_encoded = np.argmax(y_pred_drop, axis=1)
y_pred_drop_original = encoder.inverse_transform(y_pred_drop_encoded)
print(y_pred_drop_original)
# Tuning the ANN
from sklearn.model_selection import GridSearchCV
from keras.wrappers.scikit_learn import KerasClassifier
def build_network(optimizer):
network = Sequential()
network.add(Dense(input_dim = 5, units = 4, kernel_initializer = 'uniform', activation = 'relu'))
network.add(Dense(units = 4, kernel_initializer = 'uniform', activation = 'relu'))
network.add(Dense(units = 3, kernel_initializer = 'uniform', activation = 'softmax'))
network.compile(optimizer = optimizer, loss = 'categorical_crossentropy', metrics = ['accuracy'])
return network
network = KerasClassifier(build_fn = build_network, class_weight='balanced')
parameters = {'batch_size': [1, 2, 3], 'epochs': [100, 200, 300],
'optimizer': ['sgd', 'rmsprop', 'adam', 'adagrad']}
grid_search = GridSearchCV(estimator = model, param_grid = parameters, scoring = 'accuracy', cv = 5)
grid_result = grid_search.fit(X_train, np.argmax(y_train, axis=1))
best_parameters = grid_result.best_params_
best_accuracy = grid_search.best_score_
################## END ######################