跟进我上面的评论;

import tensorflow as tf

# Create a list of Pandas DataFrames
df_list = [df1, df2, ...]

# Convert the DataFrames to TensorFlow datasets
ds_list = [tf.data.Dataset.from_tensor_slices(df.values) for df in df_list]

# Concatenate the datasets
concatenated_ds = tf.data.Dataset.concatenate(ds_list)

# Apply the time series dataset function
sequence_length= m
ds = tf.data.experimental.timeseries_dataset_from_array(
    concatenated_ds,
    sequence_length=sequence_length,
    targets=concatenated_ds.skip(sequence_length),
    sequence_stride=1,
    shuffle=True)

这是一个如何转换示例然后将它们连接起来的示例。
以下是另一个示例，说明如何单独处理每个数据集，然后合并生成的数据集：

ds1 = tf.data.Dataset.from_tensor_slices(data1)
ds2 = tf.data.Dataset.from_tensor_slices(data2)
ds3 = tf.data.Dataset.from_tensor_slices(data3)

# Process dataset1
ds1 = ds1.map(preprocess_fn, num_parallel_calls=tf.data.AUTOTUNE)
ds1 = ds1.interleave(lambda x: x, cycle_length=1, num_parallel_calls=tf.data.AUTOTUNE)
ds1 = ds1.prefetch(buffer_size=tf.data.AUTOTUNE)

# Process dataset2
ds2 = ds2.map(preprocess_fn, num_parallel_calls=tf.data.AUTOTUNE)
ds2 = ds2.interleave(lambda x: x, cycle_length=1, num_parallel_calls=tf.data.AUTOTUNE)
ds2 = ds2.prefetch(buffer_size=tf.data.AUTOTUNE)

# Process dataset3
ds3 = ds3.map(preprocess_fn, num_parallel_calls=tf.data.AUTOTUNE)
ds3 = ds3.interleave(lambda x: x, cycle_length=1, num_parallel_calls=tf.data.AUTOTUNE)
ds3 = ds3.prefetch(buffer_size=tf.data.AUTOTUNE)

# Use concatenate method to combine the datasets
combined_ds = ds1.concatenate(ds2).concatenate(ds3)

另一种方法是使用tf.data.Dataset.zip（ds 1，ds 2，ds 3），它将压缩数据集并返回具有相同元素数的单个数据集。
希望这有帮助！