zhaojinghao
/
coal_materials
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
coal_materials/.ipynb_checkpoints/multi-task0102-checkpoint.i...

60 KiB
Raw Blame History

In [1]:
import os
os.environ['CUDA_DEVICE_ORDER'] = 'PCB_BUS_ID'
os.environ['CUDA_VISIBLE_DEVICES'] = '0, 1'
In [2]:
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
import matplotlib.pyplot as plt
#新增加的两行
from pylab import mpl
# 设置显示中文字体
mpl.rcParams["font.sans-serif"] = ["SimHei"]

mpl.rcParams["axes.unicode_minus"] = False
In [3]:
data_0102 = pd.read_excel('./data/20240102/20240102.xlsx', header=[0,1,2])
data_0102
Out[3]:
Unnamed: 0_level_0 氢 碳 氮 氧 弹筒发热量 挥发分 固定炭
化验编号 Had Cad Nad Oad Qb,ad Vad Fcad
Unnamed: 0_level_2 (%) (%) (%) (%) MJ/kg (%) (%)
0 2720110529 3.93 70.18 0.81 25.079 27.820 32.06 55.68
1 2720096883 3.78 68.93 0.77 26.512 27.404 29.96 54.71
2 2720109084 3.48 69.60 0.76 26.148 27.578 29.31 55.99
3 2720084708 3.47 66.71 0.76 29.055 26.338 28.58 53.87
4 2720062721 3.87 68.78 0.80 26.542 27.280 29.97 54.78
... ... ... ... ... ... ... ... ...
223 2720030490 4.12 68.85 0.97 26.055 27.864 32.94 51.89
224 2720028633 3.97 67.04 0.94 28.043 27.368 31.88 51.38
225 2720028634 4.12 68.42 0.96 26.493 27.886 33.16 52.00
226 2720017683 3.88 67.42 0.94 27.760 26.616 31.65 50.56
227 2720017678 3.81 66.74 0.92 28.530 26.688 31.02 50.82

228 rows × 8 columns

In [4]:
cols = [''.join([y for y in x if 'Unnamed' not in y]) for x in data_0102.columns]
cols
Out[4]:
['化验编号',
 '氢Had(%)',
 '碳Cad(%)',
 '氮Nad(%)',
 '氧Oad(%)',
 '弹筒发热量Qb,adMJ/kg',
 '挥发分Vad(%)',
 '固定炭Fcad(%)']
In [5]:
data_0102.columns = cols
In [6]:
data_0102
Out[6]:
化验编号 氢Had(%) 碳Cad(%) 氮Nad(%) 氧Oad(%) 弹筒发热量Qb,adMJ/kg 挥发分Vad(%) 固定炭Fcad(%)
0 2720110529 3.93 70.18 0.81 25.079 27.820 32.06 55.68
1 2720096883 3.78 68.93 0.77 26.512 27.404 29.96 54.71
2 2720109084 3.48 69.60 0.76 26.148 27.578 29.31 55.99
3 2720084708 3.47 66.71 0.76 29.055 26.338 28.58 53.87
4 2720062721 3.87 68.78 0.80 26.542 27.280 29.97 54.78
... ... ... ... ... ... ... ... ...
223 2720030490 4.12 68.85 0.97 26.055 27.864 32.94 51.89
224 2720028633 3.97 67.04 0.94 28.043 27.368 31.88 51.38
225 2720028634 4.12 68.42 0.96 26.493 27.886 33.16 52.00
226 2720017683 3.88 67.42 0.94 27.760 26.616 31.65 50.56
227 2720017678 3.81 66.74 0.92 28.530 26.688 31.02 50.82

228 rows × 8 columns

In [7]:
out_cols = ['挥发分Vad(%)', '固定炭Fcad(%)']
In [8]:
out_cols
Out[8]:
['挥发分Vad(%)', '固定炭Fcad(%)']
In [9]:
data = data_0102.copy()
In [10]:
train_data = data.dropna(subset=out_cols).fillna(0)
In [11]:
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
import tensorflow.keras.backend as K

2024-01-04 16:49:03.492957: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcudart.so.11.0
In [12]:
tf.test.is_gpu_available()

WARNING:tensorflow:From /tmp/ipykernel_45930/337460670.py:1: is_gpu_available (from tensorflow.python.framework.test_util) is deprecated and will be removed in a future version.
Instructions for updating:
Use `tf.config.list_physical_devices('GPU')` instead.
Out[12]:
False
2024-01-04 16:49:04.396035: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 AVX512F FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2024-01-04 16:49:04.407586: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcuda.so.1
2024-01-04 16:49:04.465739: E tensorflow/stream_executor/cuda/cuda_driver.cc:328] failed call to cuInit: CUDA_ERROR_INVALID_DEVICE: invalid device ordinal
2024-01-04 16:49:04.465795: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:169] retrieving CUDA diagnostic information for host: zhaojh-yv621
2024-01-04 16:49:04.465807: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:176] hostname: zhaojh-yv621
2024-01-04 16:49:04.466010: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:200] libcuda reported version is: 520.61.5
2024-01-04 16:49:04.466041: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:204] kernel reported version is: 520.61.5
2024-01-04 16:49:04.466045: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:310] kernel version seems to match DSO: 520.61.5
In [13]:
class TransformerBlock(layers.Layer):
    def __init__(self, embed_dim, num_heads, ff_dim, name, rate=0.1):
        super().__init__()
        self.att = layers.MultiHeadAttention(num_heads=num_heads, key_dim=embed_dim, name=name)
        self.ffn = keras.Sequential(
            [layers.Dense(ff_dim, activation="relu"), layers.Dense(embed_dim),]
        )
        self.layernorm1 = layers.LayerNormalization(epsilon=1e-6)
        self.layernorm2 = layers.LayerNormalization(epsilon=1e-6)
        self.dropout1 = layers.Dropout(rate)
        self.dropout2 = layers.Dropout(rate)

    def call(self, inputs, training):
        attn_output = self.att(inputs, inputs)
        attn_output = self.dropout1(attn_output, training=training)
        out1 = self.layernorm1(inputs + attn_output)
        ffn_output = self.ffn(out1)
        ffn_output = self.dropout2(ffn_output, training=training)
        return self.layernorm2(out1 + ffn_output)
In [14]:
from tensorflow.keras import Model
In [15]:
from tensorflow.keras.initializers import Constant
In [16]:
# Custom loss layer
class CustomMultiLossLayer(layers.Layer):
    def __init__(self, nb_outputs=2, **kwargs):
        self.nb_outputs = nb_outputs
        self.is_placeholder = True
        super(CustomMultiLossLayer, self).__init__(**kwargs)
        
    def build(self, input_shape=None):
        # initialise log_vars
        self.log_vars = []
        for i in range(self.nb_outputs):
            self.log_vars += [self.add_weight(name='log_var' + str(i), shape=(1,),
                                              initializer=tf.initializers.he_normal(), trainable=True)]
        super(CustomMultiLossLayer, self).build(input_shape)

    def multi_loss(self, ys_true, ys_pred):
        assert len(ys_true) == self.nb_outputs and len(ys_pred) == self.nb_outputs
        loss = 0
        for y_true, y_pred, log_var in zip(ys_true, ys_pred, self.log_vars):
            mse = (y_true - y_pred) ** 2.
            pre = K.exp(-log_var[0])
            loss += tf.abs(tf.reduce_logsumexp(pre * mse + log_var[0], axis=-1))
        return K.mean(loss)

    def call(self, inputs):
        ys_true = inputs[:self.nb_outputs]
        ys_pred = inputs[self.nb_outputs:]
        loss = self.multi_loss(ys_true, ys_pred)
        self.add_loss(loss, inputs=inputs)
        # We won't actually use the output.
        return K.concatenate(inputs, -1)
In [17]:
num_heads, ff_dim = 3, 16
In [18]:
def get_prediction_model():
    def build_output(out, out_name):
        self_block = TransformerBlock(64, num_heads, ff_dim, name=f'{out_name}_attn')
        out = self_block(out)
        out = layers.GlobalAveragePooling1D()(out)
        out = layers.Dropout(0.1)(out)
        out = layers.Dense(32, activation="relu")(out)
        # out = layers.Dense(1, name=out_name, activation="sigmoid")(out)
        return out
    inputs = layers.Input(shape=(1,len(feature_cols)), name='input')
    x = layers.Conv1D(filters=64, kernel_size=1, activation='relu')(inputs)
    # x = layers.Dropout(rate=0.1)(x)
    lstm_out = layers.Bidirectional(layers.LSTM(units=64, return_sequences=True))(x)
    lstm_out = layers.Dense(128, activation='relu')(lstm_out)
    transformer_block = TransformerBlock(128, num_heads, ff_dim, name='first_attn')
    out = transformer_block(lstm_out)
    out = layers.GlobalAveragePooling1D()(out)
    out = layers.Dropout(0.1)(out)
    out = layers.Dense(64, activation='relu')(out)
    out = K.expand_dims(out, axis=1)

    bet = build_output(out, 'vad')
    mesco = build_output(out, 'fcad')

    bet = layers.Dense(1, activation='sigmoid', name='vad')(bet)
    mesco = layers.Dense(1, activation='sigmoid', name='fcad')(mesco)

    model = Model(inputs=[inputs], outputs=[bet, mesco])
    return model
In [30]:
def get_trainable_model(prediction_model):
    inputs = layers.Input(shape=(1,len(feature_cols)), name='input')
    bet, mesco = prediction_model(inputs)
    bet_real = layers.Input(shape=(1,), name='vad_real')
    mesco_real = layers.Input(shape=(1,), name='fcad_real')
    out = CustomMultiLossLayer(nb_outputs=2)([bet_real, mesco_real, bet, mesco])
    return Model([inputs, bet_real, mesco_real], out)
In [20]:
maxs = train_data.max()
mins = train_data.min()
for col in train_data.columns:
    if maxs[col] - mins[col] == 0:
        continue
    train_data[col] = (train_data[col] - mins[col]) / (maxs[col] - mins[col])
In [21]:
train_data
Out[21]:
化验编号 氢Had(%) 碳Cad(%) 氮Nad(%) 氧Oad(%) 弹筒发热量Qb,adMJ/kg 挥发分Vad(%) 固定炭Fcad(%)
0 0.996547 0.773973 0.835414 0.456522 0.171463 0.811249 0.847737 0.828147
1 0.851118 0.671233 0.799943 0.369565 0.210254 0.782038 0.674897 0.794606
2 0.981147 0.465753 0.818956 0.347826 0.200401 0.794256 0.621399 0.838866
3 0.721367 0.458904 0.736947 0.347826 0.279094 0.707183 0.561317 0.765560
4 0.487046 0.732877 0.795687 0.434783 0.211066 0.773331 0.675720 0.797026
... ... ... ... ... ... ... ... ...
223 0.143553 0.904110 0.797673 0.804348 0.197883 0.814339 0.920165 0.697095
224 0.123762 0.801370 0.746311 0.739130 0.251699 0.779510 0.832922 0.679461
225 0.123773 0.904110 0.785471 0.782609 0.209740 0.815884 0.938272 0.700899
226 0.007066 0.739726 0.757094 0.739130 0.244038 0.726705 0.813992 0.651107
227 0.007012 0.691781 0.737798 0.695652 0.264882 0.731760 0.762140 0.660097

228 rows × 8 columns

In [22]:
# feature_cols = [x for x in train_data.columns if x not in out_cols and '第二次' not in x]
feature_cols = [x for x in train_data.columns if x not in out_cols]
use_cols = feature_cols + out_cols
In [23]:
use_data = train_data.copy()
for col in use_cols:
    use_data[col] = use_data[col].astype('float32')
In [24]:
train, valid = train_test_split(use_data[use_cols], test_size=0.3, random_state=42, shuffle=True)
valid, test = train_test_split(valid, test_size=0.3, random_state=42, shuffle=True)
In [31]:
prediction_model = get_prediction_model()
trainable_model = get_trainable_model(prediction_model)
In [32]:
prediction_model.summary()

Model: "model_3"
__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
input (InputLayer)              [(None, 1, 6)]       0                                            
__________________________________________________________________________________________________
conv1d_3 (Conv1D)               (None, 1, 64)        448         input[0][0]                      
__________________________________________________________________________________________________
bidirectional_3 (Bidirectional) (None, 1, 128)       66048       conv1d_3[0][0]                   
__________________________________________________________________________________________________
dense_30 (Dense)                (None, 1, 128)       16512       bidirectional_3[0][0]            
__________________________________________________________________________________________________
transformer_block_9 (Transforme (None, 1, 128)       202640      dense_30[0][0]                   
__________________________________________________________________________________________________
global_average_pooling1d_9 (Glo (None, 128)          0           transformer_block_9[0][0]        
__________________________________________________________________________________________________
dropout_29 (Dropout)            (None, 128)          0           global_average_pooling1d_9[0][0] 
__________________________________________________________________________________________________
dense_33 (Dense)                (None, 64)           8256        dropout_29[0][0]                 
__________________________________________________________________________________________________
tf.expand_dims_3 (TFOpLambda)   (None, 1, 64)        0           dense_33[0][0]                   
__________________________________________________________________________________________________
transformer_block_10 (Transform (None, 1, 64)        52176       tf.expand_dims_3[0][0]           
__________________________________________________________________________________________________
transformer_block_11 (Transform (None, 1, 64)        52176       tf.expand_dims_3[0][0]           
__________________________________________________________________________________________________
global_average_pooling1d_10 (Gl (None, 64)           0           transformer_block_10[0][0]       
__________________________________________________________________________________________________
global_average_pooling1d_11 (Gl (None, 64)           0           transformer_block_11[0][0]       
__________________________________________________________________________________________________
dropout_32 (Dropout)            (None, 64)           0           global_average_pooling1d_10[0][0]
__________________________________________________________________________________________________
dropout_35 (Dropout)            (None, 64)           0           global_average_pooling1d_11[0][0]
__________________________________________________________________________________________________
dense_36 (Dense)                (None, 32)           2080        dropout_32[0][0]                 
__________________________________________________________________________________________________
dense_39 (Dense)                (None, 32)           2080        dropout_35[0][0]                 
__________________________________________________________________________________________________
vad (Dense)                     (None, 1)            33          dense_36[0][0]                   
__________________________________________________________________________________________________
fcad (Dense)                    (None, 1)            33          dense_39[0][0]                   
==================================================================================================
Total params: 402,482
Trainable params: 402,482
Non-trainable params: 0
__________________________________________________________________________________________________
In [33]:
from tensorflow.keras import optimizers
from tensorflow.python.keras.utils.vis_utils import plot_model
In [34]:
X = np.expand_dims(train[feature_cols].values, axis=1)
Y = [x for x in train[out_cols].values.T]
Y_valid = [x for x in valid[out_cols].values.T]
In [35]:
from keras.callbacks import ReduceLROnPlateau
reduce_lr = ReduceLROnPlateau(monitor='val_loss', patience=10, mode='auto')
In [40]:
trainable_model.compile(optimizer='adam', loss=None)
hist = trainable_model.fit([X, Y[0], Y[1]], epochs=120, batch_size=8, verbose=1, 
                           validation_data=[np.expand_dims(valid[feature_cols].values, axis=1), Y_valid[0], Y_valid[1]],
                           callbacks=[reduce_lr]
                           )

Epoch 1/120
20/20 [==============================] - 5s 59ms/step - loss: 1.8316 - val_loss: 1.8096
Epoch 2/120
20/20 [==============================] - 0s 25ms/step - loss: 1.7903 - val_loss: 1.7691
Epoch 3/120
20/20 [==============================] - 0s 26ms/step - loss: 1.7506 - val_loss: 1.7307
Epoch 4/120
20/20 [==============================] - 0s 26ms/step - loss: 1.7110 - val_loss: 1.6914
Epoch 5/120
20/20 [==============================] - 0s 21ms/step - loss: 1.6711 - val_loss: 1.6497
Epoch 6/120
20/20 [==============================] - 0s 24ms/step - loss: 1.6314 - val_loss: 1.6098
Epoch 7/120
20/20 [==============================] - 0s 24ms/step - loss: 1.5909 - val_loss: 1.5695
Epoch 8/120
20/20 [==============================] - 0s 24ms/step - loss: 1.5506 - val_loss: 1.5296
Epoch 9/120
20/20 [==============================] - 0s 24ms/step - loss: 1.5109 - val_loss: 1.4891
Epoch 10/120
20/20 [==============================] - 0s 25ms/step - loss: 1.4706 - val_loss: 1.4500
Epoch 11/120
20/20 [==============================] - 0s 25ms/step - loss: 1.4306 - val_loss: 1.4104
Epoch 12/120
20/20 [==============================] - 0s 22ms/step - loss: 1.3907 - val_loss: 1.3746
Epoch 13/120
20/20 [==============================] - 0s 23ms/step - loss: 1.3508 - val_loss: 1.3296
Epoch 14/120
20/20 [==============================] - 0s 26ms/step - loss: 1.3106 - val_loss: 1.2895
Epoch 15/120
20/20 [==============================] - 0s 26ms/step - loss: 1.2706 - val_loss: 1.2515
Epoch 16/120
20/20 [==============================] - 0s 22ms/step - loss: 1.2315 - val_loss: 1.2104
Epoch 17/120
20/20 [==============================] - 0s 23ms/step - loss: 1.1908 - val_loss: 1.1702
Epoch 18/120
20/20 [==============================] - 0s 25ms/step - loss: 1.1508 - val_loss: 1.1320
Epoch 19/120
20/20 [==============================] - 0s 26ms/step - loss: 1.1114 - val_loss: 1.0917
Epoch 20/120
20/20 [==============================] - 0s 24ms/step - loss: 1.0718 - val_loss: 1.0513
Epoch 21/120
20/20 [==============================] - 0s 23ms/step - loss: 1.0315 - val_loss: 1.0178
Epoch 22/120
20/20 [==============================] - 0s 24ms/step - loss: 0.9918 - val_loss: 0.9704
Epoch 23/120
20/20 [==============================] - 0s 23ms/step - loss: 0.9511 - val_loss: 0.9321
Epoch 24/120
20/20 [==============================] - 0s 26ms/step - loss: 0.9114 - val_loss: 0.8913
Epoch 25/120
20/20 [==============================] - 0s 26ms/step - loss: 0.8718 - val_loss: 0.8520
Epoch 26/120
20/20 [==============================] - 0s 25ms/step - loss: 0.8314 - val_loss: 0.8124
Epoch 27/120
20/20 [==============================] - 0s 26ms/step - loss: 0.7922 - val_loss: 0.7727
Epoch 28/120
20/20 [==============================] - 0s 25ms/step - loss: 0.7519 - val_loss: 0.7307
Epoch 29/120
20/20 [==============================] - 0s 26ms/step - loss: 0.7119 - val_loss: 0.6932
Epoch 30/120
20/20 [==============================] - 0s 22ms/step - loss: 0.6720 - val_loss: 0.6531
Epoch 31/120
20/20 [==============================] - 0s 23ms/step - loss: 0.6336 - val_loss: 0.6155
Epoch 32/120
20/20 [==============================] - 1s 26ms/step - loss: 0.5931 - val_loss: 0.5738
Epoch 33/120
20/20 [==============================] - 0s 26ms/step - loss: 0.5517 - val_loss: 0.5324
Epoch 34/120
20/20 [==============================] - 0s 26ms/step - loss: 0.5135 - val_loss: 0.4943
Epoch 35/120
20/20 [==============================] - 0s 21ms/step - loss: 0.4724 - val_loss: 0.4602
Epoch 36/120
20/20 [==============================] - 0s 23ms/step - loss: 0.4326 - val_loss: 0.4126
Epoch 37/120
20/20 [==============================] - 0s 26ms/step - loss: 0.3947 - val_loss: 0.3758
Epoch 38/120
20/20 [==============================] - 0s 25ms/step - loss: 0.3558 - val_loss: 0.3350
Epoch 39/120
20/20 [==============================] - 0s 22ms/step - loss: 0.3154 - val_loss: 0.3031
Epoch 40/120
20/20 [==============================] - 0s 23ms/step - loss: 0.2771 - val_loss: 0.2592
Epoch 41/120
20/20 [==============================] - 0s 25ms/step - loss: 0.2459 - val_loss: 0.2370
Epoch 42/120
20/20 [==============================] - 1s 27ms/step - loss: 0.2267 - val_loss: 0.2210
Epoch 43/120
20/20 [==============================] - 0s 24ms/step - loss: 0.2050 - val_loss: 0.1947
Epoch 44/120
20/20 [==============================] - 0s 26ms/step - loss: 0.1840 - val_loss: 0.1728
Epoch 45/120
20/20 [==============================] - 0s 21ms/step - loss: 0.1628 - val_loss: 0.1533
Epoch 46/120
20/20 [==============================] - 0s 23ms/step - loss: 0.1430 - val_loss: 0.1322
Epoch 47/120
20/20 [==============================] - 0s 23ms/step - loss: 0.1230 - val_loss: 0.1147
Epoch 48/120
20/20 [==============================] - 1s 24ms/step - loss: 0.1026 - val_loss: 0.0940
Epoch 49/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0830 - val_loss: 0.0750
Epoch 50/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0639 - val_loss: 0.0529
Epoch 51/120
20/20 [==============================] - 0s 22ms/step - loss: 0.0436 - val_loss: 0.0352
Epoch 52/120
20/20 [==============================] - 0s 24ms/step - loss: 0.0241 - val_loss: 0.0162
Epoch 53/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0092 - val_loss: 0.0084
Epoch 54/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0067 - val_loss: 0.0074
Epoch 55/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0080 - val_loss: 0.0071
Epoch 56/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0070 - val_loss: 0.0063
Epoch 57/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0062 - val_loss: 0.0076
Epoch 58/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0056 - val_loss: 0.0048
Epoch 59/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0050 - val_loss: 0.0071
Epoch 60/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0057 - val_loss: 0.0054
Epoch 61/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0044 - val_loss: 0.0092
Epoch 62/120
20/20 [==============================] - 1s 26ms/step - loss: 0.0068 - val_loss: 0.0070
Epoch 63/120
20/20 [==============================] - 1s 24ms/step - loss: 0.0059 - val_loss: 0.0065
Epoch 64/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0055 - val_loss: 0.0060
Epoch 65/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0053 - val_loss: 0.0056
Epoch 66/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0058 - val_loss: 0.0077
Epoch 67/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0051 - val_loss: 0.0054
Epoch 68/120
20/20 [==============================] - 0s 21ms/step - loss: 0.0047 - val_loss: 0.0048
Epoch 69/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0041 - val_loss: 0.0048
Epoch 70/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0037 - val_loss: 0.0049
Epoch 71/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0041 - val_loss: 0.0049
Epoch 72/120
20/20 [==============================] - 0s 21ms/step - loss: 0.0036 - val_loss: 0.0049
Epoch 73/120
20/20 [==============================] - 1s 24ms/step - loss: 0.0038 - val_loss: 0.0048
Epoch 74/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0037 - val_loss: 0.0050
Epoch 75/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0037 - val_loss: 0.0048
Epoch 76/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0038 - val_loss: 0.0048
Epoch 77/120
20/20 [==============================] - 0s 21ms/step - loss: 0.0037 - val_loss: 0.0048
Epoch 78/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0038 - val_loss: 0.0048
Epoch 79/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0036 - val_loss: 0.0048
Epoch 80/120
20/20 [==============================] - 0s 22ms/step - loss: 0.0034 - val_loss: 0.0048
Epoch 81/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0034 - val_loss: 0.0047
Epoch 82/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0037 - val_loss: 0.0047
Epoch 83/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0033 - val_loss: 0.0047
Epoch 84/120
20/20 [==============================] - 0s 22ms/step - loss: 0.0037 - val_loss: 0.0047
Epoch 85/120
20/20 [==============================] - 0s 24ms/step - loss: 0.0036 - val_loss: 0.0047
Epoch 86/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0034 - val_loss: 0.0047
Epoch 87/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0034 - val_loss: 0.0047
Epoch 88/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0034 - val_loss: 0.0047
Epoch 89/120
20/20 [==============================] - 0s 21ms/step - loss: 0.0036 - val_loss: 0.0047
Epoch 90/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0036 - val_loss: 0.0047
Epoch 91/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0036 - val_loss: 0.0047
Epoch 92/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0034 - val_loss: 0.0047
Epoch 93/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0034 - val_loss: 0.0047
Epoch 94/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0037 - val_loss: 0.0047
Epoch 95/120
20/20 [==============================] - 0s 21ms/step - loss: 0.0035 - val_loss: 0.0047
Epoch 96/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0032 - val_loss: 0.0047
Epoch 97/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0035 - val_loss: 0.0047
Epoch 98/120
20/20 [==============================] - 0s 21ms/step - loss: 0.0038 - val_loss: 0.0047
Epoch 99/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0033 - val_loss: 0.0047
Epoch 100/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0036 - val_loss: 0.0047
Epoch 101/120
20/20 [==============================] - 1s 26ms/step - loss: 0.0033 - val_loss: 0.0047
Epoch 102/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0035 - val_loss: 0.0047
Epoch 103/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0035 - val_loss: 0.0047
Epoch 104/120
20/20 [==============================] - 0s 22ms/step - loss: 0.0035 - val_loss: 0.0047
Epoch 105/120
20/20 [==============================] - 0s 23ms/step - loss: 0.0036 - val_loss: 0.0047
Epoch 106/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0034 - val_loss: 0.0047
Epoch 107/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0036 - val_loss: 0.0047
Epoch 108/120
20/20 [==============================] - 0s 21ms/step - loss: 0.0034 - val_loss: 0.0047
Epoch 109/120
20/20 [==============================] - 1s 24ms/step - loss: 0.0037 - val_loss: 0.0047
Epoch 110/120
20/20 [==============================] - 0s 24ms/step - loss: 0.0038 - val_loss: 0.0047
Epoch 111/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0036 - val_loss: 0.0047
Epoch 112/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0035 - val_loss: 0.0047
Epoch 113/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0037 - val_loss: 0.0047
Epoch 114/120
20/20 [==============================] - 0s 20ms/step - loss: 0.0035 - val_loss: 0.0047
Epoch 115/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0033 - val_loss: 0.0047
Epoch 116/120
20/20 [==============================] - 0s 21ms/step - loss: 0.0036 - val_loss: 0.0047
Epoch 117/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0032 - val_loss: 0.0047
Epoch 118/120
20/20 [==============================] - 0s 26ms/step - loss: 0.0036 - val_loss: 0.0047
Epoch 119/120
20/20 [==============================] - 0s 25ms/step - loss: 0.0037 - val_loss: 0.0047
Epoch 120/120
20/20 [==============================] - 0s 21ms/step - loss: 0.0036 - val_loss: 0.0047
In [41]:
rst = prediction_model.predict(np.expand_dims(test[feature_cols], axis=1))
rst
Out[41]:
[array([[0.73740077],
        [0.89292204],
        [0.7599046 ],
        [0.67802393],
        [0.6815233 ],
        [0.88627005],
        [0.6121343 ],
        [0.7072234 ],
        [0.8561135 ],
        [0.52762157],
        [0.8325021 ],
        [0.50241977],
        [0.8242289 ],
        [0.68957335],
        [0.6980361 ],
        [0.82116604],
        [0.8566438 ],
        [0.53687835],
        [0.56832707],
        [0.78476715],
        [0.85638577]], dtype=float32),
 array([[0.68600863],
        [0.78454906],
        [0.8179163 ],
        [0.94351083],
        [0.86383885],
        [0.69705516],
        [0.6913491 ],
        [0.80277354],
        [0.93557894],
        [0.82278305],
        [0.82674253],
        [0.93518937],
        [0.8094449 ],
        [0.9206344 ],
        [0.7747319 ],
        [0.9137207 ],
        [0.9491073 ],
        [0.93225   ],
        [0.6185102 ],
        [0.8867341 ],
        [0.82890105]], dtype=float32)]
In [42]:
[np.exp(K.get_value(log_var[0]))**0.5 for log_var in trainable_model.layers[-1].log_vars]
Out[42]:
[0.9991559102070927, 0.9998196796918477]
In [43]:
pred_rst = pd.DataFrame.from_records(np.squeeze(np.asarray(rst), axis=2).T, columns=out_cols)
In [44]:
real_rst = test[out_cols].copy()
In [45]:
for col in out_cols:
    pred_rst[col] = pred_rst[col] * (maxs[col] - mins[col]) + mins[col]
    real_rst[col] = real_rst[col] * (maxs[col] - mins[col]) + mins[col]
In [46]:
real_rst.columns
Out[46]:
Index(['挥发分Vad(%)', '固定炭Fcad(%)'], dtype='object')
In [47]:
y_pred_vad = pred_rst['挥发分Vad(%)'].values.reshape(-1,)
y_pred_fcad = pred_rst['固定炭Fcad(%)'].values.reshape(-1,)
y_true_vad = real_rst['挥发分Vad(%)'].values.reshape(-1,)
y_true_fcad = real_rst['固定炭Fcad(%)'].values.reshape(-1,)
In [48]:
from sklearn.metrics import mean_squared_error, mean_absolute_error, r2_score, mean_absolute_percentage_error
In [49]:
def print_eva(y_true, y_pred, tp):
    MSE = mean_squared_error(y_true, y_pred)
    RMSE = np.sqrt(MSE)
    MAE = mean_absolute_error(y_true, y_pred)
    MAPE = mean_absolute_percentage_error(y_true, y_pred)
    R_2 = r2_score(y_true, y_pred)
    print(f"COL: {tp}, MSE: {format(MSE, '.2E')}", end=',')
    print(f'RMSE: {round(RMSE, 4)}', end=',')
    print(f'MAPE: {round(MAPE, 4) * 100} %', end=',')
    print(f'MAE: {round(MAE, 4)}', end=',')
    print(f'R_2: {round(R_2, 4)}')
    return [MSE, RMSE, MAE, MAPE, R_2]
In [51]:
pm25_eva = print_eva(y_true_vad, y_pred_vad, tp='挥发分Vad')
pm10_eva = print_eva(y_true_fcad, y_pred_fcad, tp='固定炭Fcad')

COL: 比表面积, MSE: 3.35E-01,RMSE: 0.5791,MAPE: 1.6400000000000001 %,MAE: 0.5041,R_2: 0.8698
COL: 总孔体积, MSE: 1.11E+00,RMSE: 1.0549,MAPE: 1.5 %,MAE: 0.8137,R_2: 0.876
In [ ]:
In [ ]: