markilue
/
self_example
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

leetcode更新

This commit is contained in:
kevinding1125 2023-06-22 10:23:39 +08:00
parent a35ccc2f63
commit aa79b3d094
2 changed files with 24 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

46
TensorFlow_eaxmple/Model_train_test/RUL/test/LSTMContinueTest.py
View File

@ -25,7 +25,7 @@ from pylab import *
''' '''
hidden_num = 10 # LSTM细胞个数 hidden_num = 10 # LSTM细胞个数
feature = 10 # 一个点的维度 feature = 10 # 一个点的维度
batch_size = 32 batch_size = 8
EPOCH = 1000 EPOCH = 1000
unit = 512 # LSTM的维度 unit = 512 # LSTM的维度
predict_num = 50 # 预测个数 predict_num = 50 # 预测个数
@ -105,8 +105,6 @@ train_data.shape: (total_dims - filter_num - 1, filter_num,dims) :(570,600,30)
predict_data.shape: (total_dims - filter_num, filter_num) :(571,600,30) predict_data.shape: (total_dims - filter_num, filter_num) :(571,600,30)
train_label.shape: (total_dims - filter_num - 1, filter_num) :(570,600) train_label.shape: (total_dims - filter_num - 1, filter_num) :(570,600)
''' '''
def remove(train_data, train_label, batch_size): def remove(train_data, train_label, batch_size):
epoch, _, _ = train_data.shape epoch, _, _ = train_data.shape
size = int(epoch / batch_size) size = int(epoch / batch_size)
@ -118,8 +116,6 @@ train_data.shape: (1230, 10, 10)
train_label.shape: (1230, 10) train_label.shape: (1230, 10)
train_label_single.shape: (1230,) train_label_single.shape: (1230,)
''' '''
def splitValData(data, label, label_single, predict_num=50): def splitValData(data, label, label_single, predict_num=50):
sample, hidden, feature = data.shape sample, hidden, feature = data.shape
@ -139,7 +135,7 @@ def predict_model_multi(filter_num, dims):
tf.config.experimental_run_functions_eagerly(True) tf.config.experimental_run_functions_eagerly(True)
input = tf.keras.Input(shape=[filter_num, dims]) input = tf.keras.Input(shape=[filter_num, dims])
input = tf.cast(input, tf.float32) # input = tf.cast(input, tf.float32)
#### 官方 #### 官方
# LSTM = tf.keras.layers.LSTM(units=512, return_sequences=True)(input) # LSTM = tf.keras.layers.LSTM(units=512, return_sequences=True)(input)
@ -210,31 +206,31 @@ if __name__ == '__main__':
train_label, train_label,
train_label_single, train_label_single,
predict_num=predict_num) predict_num=predict_num)
# #### TODO 训练 # # #### TODO 训练
model = predict_model_multi(hidden_num, feature) # model = predict_model_multi(hidden_num, feature)
checkpoint = tf.keras.callbacks.ModelCheckpoint( # checkpoint = tf.keras.callbacks.ModelCheckpoint(
filepath=save_name, # filepath=save_name,
monitor='val_loss', # monitor='val_loss',
verbose=2, # verbose=2,
save_best_only=True, # save_best_only=True,
mode='min') # mode='min')
lr_scheduler = tf.keras.callbacks.ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=20, min_lr=0.001) # lr_scheduler = tf.keras.callbacks.ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=20, min_lr=0.001)
#
model.compile(optimizer=tf.optimizers.SGD(), loss=tf.losses.mse) # model.compile(optimizer=tf.optimizers.SGD(), loss=tf.losses.mse)
# model.compile(optimizer=tf.optimizers.SGD(learning_rate=0.001), loss=FTMSE()) # # model.compile(optimizer=tf.optimizers.SGD(learning_rate=0.001), loss=FTMSE())
model.summary() # model.summary()
early_stop = EarlyStopping(monitor='val_loss', min_delta=0.0001, patience=100, mode='min', verbose=1) # early_stop = EarlyStopping(monitor='val_loss', min_delta=0.0001, patience=100, mode='min', verbose=1)
#
history = model.fit(train_data, train_label, epochs=EPOCH, # history = model.fit(train_data, train_label, epochs=EPOCH,
batch_size=batch_size, validation_data=(val_data, val_label_single), shuffle=True, verbose=2, # batch_size=batch_size, validation_data=(val_data, val_label_single), shuffle=True, verbose=2,
callbacks=[checkpoint, lr_scheduler, early_stop]) # callbacks=[checkpoint, lr_scheduler, early_stop])
#### TODO 测试 #### TODO 测试
# trained_model = tf.keras.models.load_model(save_name, custom_objects={'LSTMLayer': LSTMLayer, 'FTMSE': FTMSE}) # trained_model = tf.keras.models.load_model(save_name, custom_objects={'LSTMLayer': LSTMLayer, 'FTMSE': FTMSE})
# todo 解决自定义loss无法导入的问题 # todo 解决自定义loss无法导入的问题
trained_model = tf.keras.models.load_model(save_name, compile=False, custom_objects={'LSTMLayer': LSTMLayer,'DCTChannelAttention':DCTChannelAttention}) trained_model = tf.keras.models.load_model(save_name, compile=False, custom_objects={'LSTMLayer': LSTMLayer,'LightChannelAttention1':LightChannelAttention})
trained_model.compile(optimizer=tf.optimizers.SGD(), loss=FTMSE()) trained_model.compile(optimizer=tf.optimizers.SGD(), loss=FTMSE())
# 使用已知的点进行预测 # 使用已知的点进行预测

6
TensorFlow_eaxmple/Model_train_test/model/ChannelAttention/Light_channelAttention.py
View File

@ -86,7 +86,6 @@ class LightChannelAttention1(layers.Layer):
raise ValueError('Inputs to `DynamicChannelAttention` should have rank 3. ' raise ValueError('Inputs to `DynamicChannelAttention` should have rank 3. '
'Received input shape:', str(input_shape)) 'Received input shape:', str(input_shape))
print(input_shape)
# GAP # GAP
self.GAP = tf.keras.layers.GlobalAvgPool1D() self.GAP = tf.keras.layers.GlobalAvgPool1D()
self.c1 = tf.keras.layers.Conv1D(filters=input_shape[2], kernel_size=1, padding='SAME') self.c1 = tf.keras.layers.Conv1D(filters=input_shape[2], kernel_size=1, padding='SAME')
@ -98,13 +97,14 @@ class LightChannelAttention1(layers.Layer):
# GAP # GAP
GAP = self.GAP(DWC1) GAP = self.GAP(DWC1)
GAP = tf.expand_dims(GAP, axis=1) GAP = tf.expand_dims(GAP, axis=1)
c1 = self.c1(GAP) c1 = self.c1(GAP)
c1 = tf.keras.layers.BatchNormalization()(c1) c1 = tf.keras.layers.BatchNormalization()(c1)
s1 = tf.nn.sigmoid(c1) s1 = tf.nn.sigmoid(c1)
print(s1)
s1 = tf.broadcast_to(s1, [batch_size, length, channel])
s1 = tf.tile(s1, [1, length, 1])
return s1 * inputs return s1 * inputs