tensorflow矩阵扩充

a 的形状为([3,1,3,3,3])
需要在维度axis=1上扩充a至c=[3,8,3,3,3]
由于tensorflow的矩阵在某一维度做切片比较麻烦，采用了如下方法
a = 10*tf.random_normal([3,3,3,3])
b = tf.expand_dims(a,axis=1)
e = b
for i in range(8-1):
e = tf.concat([e,b],axis=1)
with tf.Session() as sess:
aa = sess.run(e)
print(aa.shape)
输出为[3,8,3,3,3]