按你说的修改后出现ValueError: Layer fs_feature weight shape (3, 1, 1) is not compatible with provided weight shape (1, 3).

非常抱歉，我之前的回答有误。在修改第一层权重的代码中，应该使用np.array([weights]).T而不是np.transpose(np.array([weights]))。修改后的代码如下：

import numpy as np
from tensorflow import keras
from tensorflow.keras.layers import Conv1D

data = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9])
data = data.reshape(1, -1, 1)
print(data.shape)

input = keras.Input(shape=(9, 1))
output = Conv1D(filters=1, kernel_size=3, padding='causal', activation='linear')(input)
model = keras.Model(inputs=input, outputs=output, name="fs_feature")
model.summary()

weights_list = model.get_weights()
print("权重列表：", weights_list)

weights = np.ones(3) / 3
weights_list[0] = np.array([weights]).T
model.set_weights(weights_list)

result = model.predict(data)
print("原始数据：", data)
print("滑动平均结果：", result.squeeze())

请注意，这里的滑动平均是通过卷积实现的。如果你想实现其他类型的滑动平均，请根据具体需求修改代码。