LSTM时间步骤？

Question

每个人的下午好，

我是编程的新手，所以很抱歉，如果我对问题的愚蠢评论，

我正在建立一个LSTM模型，其中我有4个不同的输入，例如空气临时，空气湿度，风速等。我需要产生风力预测的输出。

我的数据集为一年，时间步长为1小时。我首先在制作LSTM时挣扎，因为我需要输入维度为3D Andi在网上找到了解决方案并应用了解决方案。但是现在，我需要更改时间步骤，以查看模型中表现更好的情况，但是我对如何更改它一无所知。

我的代码看起来像：

import tensorflow as tf
import csv
from sklearn.model_selection import train_test_split
import numpy as np
import matplotlib.pyplot as plt
import time

#Read DATA
with open("full_year_data.csv") as f:
    reader = csv.reader(f)
    next(reader)
    data = []
    for row in reader:
        data.append({
            "inputs": [float(cell) for cell in row[:4]],
            "output": [float(cell) for cell in row[4:]]
        })

#separate data into training and testing groups
inputs = [row["inputs"] for row in data]
inputs = np.array(inputs) #shape = (743, 4)
output = [row["output"] for row in data]
output = np.array(output) #shape= (743,1)

X_training, X_testing, y_training, y_testing = train_test_split(
    inputs, output, test_size=0.25
)


X_training = np.expand_dims(X_training,1)
X_testing = np.expand_dims(X_testing,1)

nodes_hidden = 
Learning_algorithm = tf.keras.optimizers.Adam()

model2 = tf.keras.models.Sequential()
model2.add(tf.keras.layers.LSTM(units = nodes_hidden, return_sequences =True, input_shape=X_training.shape[1:], activation='relu'))
model2.add(tf.keras.layers.LSTM(units = nodes_hidden, return_sequences =True, activation='relu'))
model2.add(tf.keras.layers.LSTM(units = nodes_hidden, return_sequences =True, activation='relu'))
model2.add(tf.keras.layers.LSTM(units = nodes_hidden, return_sequences =True, activation='relu'))
model2.add(tf.keras.layers.Dense(1, activation='sigmoid'))

model2.compile(optimizer = 'adam', loss = 'mean_squared_error')

model2.fit(X_training, y_training, epochs=100, batch_size=0)
model2.evaluate(X_testing, y_testing, verbose=2)

我正在使用TensorFlow和Keras制作LSTM模型，关于如何更改我的时间步骤的任何帮助，将非常感谢

您的时间！