多层感知机、循环神经网络、Stateful循环神经网络、Stateful叠加循环神经网络预测数值

最新推荐文章于 2024-12-06 13:59:54 发布

MAOX789

最新推荐文章于 2024-12-06 13:59:54 发布

阅读量351

点赞数

分类专栏：深度学习文章标签：神经网络 tensorflow 机器学习深度学习

本文链接：https://blog.youkuaiyun.com/MXIAO56789/article/details/118941901

版权

深度学习专栏收录该内容

13 篇文章

订阅专栏

该博客介绍了如何使用多层感知器和LSTM循环神经网络进行时间序列预测。首先，通过创建数据集并进行预处理，然后构建和训练多层感知器模型。接着，展示了LSTM模型的训练过程，包括Stateful LSTM的实现。最后，对模型的性能进行了评估，并展示了预测结果。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

在这里插入图片描述

#多层感知器神经网络模型
# 0. 调用要使用的包
import numpy as np
from keras.models import Sequential
from keras.layers import Dense, LSTM, Dropout
from sklearn.preprocessing import MinMaxScaler
import matplotlib.pyplot as plt
%matplotlib inline
def create_dataset(signal_data, look_back=1):
    dataX, dataY = [], []
    for i in range(len(signal_data)-look_back):
        dataX.append(signal_data[i:(i+look_back), 0])
        dataY.append(signal_data[i + look_back, 0])
    return np.array(dataX), np.array(dataY)
look_back = 40
# 1. 生成数据集
signal_data = 3 * np.sin(np.arange(4000)*(20*np.pi/1000))[:,None]
# 数据预处理
plot_x = np.arange(4000)
plot_y = signal_data
plt.plot(plot_x,plot_y)
plt.show

scaler = MinMaxScaler(feature_range=(0, 1))
signal_data = scaler.fit_transform(signal_data)
# 分离数据
train = signal_data[0:2400]
val = signal_data[2400:3600]
test = signal_data[3600:]
# 生成数据集
x_train, y_train = create_dataset(train, look_back)
x_val, y_val = create_dataset(val, look_back)
x_test, y_test = create_dataset(test, look_back)
# 数据集预处理
x_train = np.reshape(x_train, (x_train.shape[0], x_train.shape[1], 1))
x_val = np.reshape(x_val, (x_val.shape[0], x_val.shape[1], 1))
x_test = np.reshape(x_test, (x_test.shape[0], x_test.shape[1], 1))
x_train = np.squeeze(x_train)
x_val = np.squeeze(x_val)
x_test = np.squeeze(x_test)
# 2. 搭建模型
model = Sequential()
model.add(Dense(32,input_dim=40,activation="relu"))
model.add(Dropout(0.3))
for i in range(2):
    model.add(Dense(32,activation="relu"))
    model.add(Dropout(0.3))
model.add(Dense(1))
# 3. 设置模型训练过程
model.compile(loss='mean_squared_error', optimizer='adagrad')
# 4. 训练模型
hist = model.fit(x_train, y_train, epochs=400, batch_size=64, validation_data=(x_val, y_val))

# 5. 查看训练过程
plt.plot(hist.history['loss'])
plt.plot(hist.history['val_loss'])
plt.ylim(0.0, 0.3)
plt.ylabel('loss')
plt.xlabel('epoch')
plt.legend(['train', 'val'], loc='upper right')
plt.show()
# 6. 评价模型
trainScore = model.evaluate(x_train, y_train, verbose=0)
print('Train Score: ', trainScore)
valScore = model.evaluate(x_val, y_val, verbose=0)
print('Validation Score: ', valScore)
testScore = model.evaluate(x_test, y_test, verbose=0)
print('Test Score ', testScore)

# 7. 使用模型
look_ahead = 360       #预测前300个
xhat = x_test[0, None]
predictions = np.zeros((look_ahead,1))    #初始化为0
for i in range(look_ahead):
    prediction = model.predict(xhat, batch_size=32)
    predictions[i] = prediction
    xhat = np.hstack([xhat[:,1:],prediction])
plt.figure(figsize=(12,6))
plt.plot(np.arange(look_ahead),predictions,'r',label="prediction")
plt.plot(np.arange(look_ahead),y_test[:look_ahead],'black',label="test function")
plt.legend()
plt.show()

# 循环神经网络模型
model = Sequential()
model.add(LSTM(32, input_shape=(None, 1)))
model.add(Dropout(0.3))
model.add(Dense(1))

# Stateful 循环神经网络模型
for i in range(40):
    model.fit(x_train, y_train, epochs=1, batch_size=1, shuffle=False, callbacks=[custom_hist], validation_data=(x_val, y_val))
    model.reset_states()  #删除记忆

#Stateful 叠加循环神经网络模型
model = Sequential()
for i in range(2):
    model.add(LSTM(32, batch_input_shape=(1, look_back, 1), stateful=True, return_sequences=True))
    model.add(Dropout(0.3))
model.add(LSTM(32, batch_input_shape=(1, look_back, 1), stateful=True))
model.add(Dropout(0.3))
model.add(Dense(1))