lstm tensorflow,keras的实现

最新推荐文章于 2022-04-10 12:00:25 发布

无敌的绅士

最新推荐文章于 2022-04-10 12:00:25 发布

阅读量788

点赞数

分类专栏：深度学习 lstm 文章标签： tensorflow python 深度学习神经网络

本文链接：https://blog.csdn.net/weixin_42164407/article/details/112311487

版权

深度学习同时被 2 个专栏收录

7 篇文章

订阅专栏

lstm

3 篇文章

订阅专栏

这篇博客介绍了如何在Python中利用TensorFlow实现LSTM模型，用于预测二维坐标序列中的第五个点。首先，从Excel文件读取数据并预处理，然后创建时间序列数据集。接着，构建LSTM模型并训练，最后保存模型权重，并用模型进行预测，绘制预测结果与实际值的对比图表。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

lstm的tensorflow实现

多变量输入（通过输入五个点(二维,x,y)，预测第十个点）,我是在jupyter notebook上实现的，如果想在pycharm上运行，需要稍作修改。
读取的excel文件是n行2列的连续点，另外，还有python和c++的lstm实现（通过加载训练好的权重进行预测）在我其他博客里，整个项目可以去我github下载。
整个的测试集的预测效果

#!/usr/bin/env python
# coding: utf-8

# In[108]:


import numpy
import matplotlib.pyplot as plt
from pandas import read_csv,read_excel
import math
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense
from tensorflow.keras.layers import LSTM
from tensorflow.keras.layers import TimeDistributed
from sklearn.preprocessing import MinMaxScaler
from sklearn.metrics import mean_squared_error
import tensorflow as tf
import os
#os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"
#os.environ["CUDA_VISIBLE_DEVICES"] = "-1"


# In[109]:


dataframe = read_excel('./data-lstm/lstm1.xlsx', usecols=[0,1], skipfooter=3)
dataset = dataframe.values
# 将整型变为float
dataset = dataset.astype('float32')


# In[110]:


def create_dataset(dataset, look_back=1):
    dataX, dataY = [], []
    predict_len = 5
    for i in range(len(dataset)-look_back-1 -predict_len):
        a = dataset[i:(i+look_back), 0:2]
        dataX.append(a)
        dataY.append(dataset[i +look_back + predict_len , 0:2])
    return numpy.array(dataX), numpy.array(dataY)


# In[111]:


# fix random seed for reproducibility
numpy.random.seed(7)

scaler = MinMaxScaler(feature_range=(0, 1))
dataset = scaler.fit_transform(dataset)

# split into train and test sets
train_size = int(len(dataset) * 0.67)
test_size = len(dataset) - train_size
train, test = dataset[0:train_size,:2], dataset[train_size:len(dataset),:2]

# use this function to prepare the train and test datasets for modeling
look_back = 5
trainX, trainY = create_dataset(train, look_back)
testX, testY = create_dataset(test, look_back)


# In[100]:


# create and fit the LSTM network
model = Sequential()
model.add(LSTM(10, input_shape=(look_back,2)))
model.add(Dense(2))
model.compile(loss='mean_squared_error', optimizer='adam')


# In[101]:


print(trainX.shape)
#trainX = numpy.reshape(trainX, (trainX.shape[0], trainX.shape[1]),2)
#testX = numpy.reshape(testX, (testX.shape[0],  testX.shape[1]),2)
#print(trainX.shape)


# In[112]:


#model =  tf.keras.models.load_model(r'C:\Users\gj7520\Desktop\pythob_files\lstm\lstm2\lstm1_best2.h5')


# In[102]:



callbacks = [
    tf.keras.callbacks.ModelCheckpoint('./models/lstm.pbtxt',
                                        monitor='loss',
                                        save_best_only=True)]
model.fit(trainX, trainY, epochs=50, batch_size=1, verbose=2)
model.save(r'C:\Users\gj7520\Desktop\pythob_files\lstm\lstm2\lstm1.h5')
model.save('./data-lstm/lstm2.pbtxt')


# In[103]:


model.save_weights('lstm.ckpt')


# In[113]:


# make predictions
trainPredict = model.predict(trainX)
testPredict = model.predict(testX)

# invert predictions
trainPredict = scaler.inverse_transform(trainPredict)

trainY = scaler.inverse_transform(trainY)

testPredict = scaler.inverse_transform(testPredict)

testY = scaler.inverse_transform(testY)


# In[114]:


plt.plot(testY)
plt.plot(testPredict)
plt.show()


# In[115]:


# shift train predictions for plotting
predict_len=5
trainPredictPlot = numpy.empty_like(dataset)
trainPredictPlot[:, :] = numpy.nan
trainPredictPlot[look_back:len(trainPredict)+predict_len, :] = trainPredict

# shift test predictions for plotting
testPredictPlot = numpy.empty_like(dataset)
testPredictPlot[:, :] = numpy.nan
testPredictPlot[len(trainPredict)+(look_back*2)+(predict_len*2)+1:len(dataset)-1, :] = testPredict


# In[116]:


# plot baseline and predictions
plt.plot(scaler.inverse_transform(dataset))
#plt.plot(dataset)
plt.plot(trainPredictPlot)
plt.plot(testPredictPlot)
plt.show()