您可以转换为秒（或毫秒或微秒）并返回到时间戳：

select timestamp_seconds(cast(avg(unix_seconds(timestamp)) as int64))
from my_table

下面是bigquery标准sql
我假设你想知道用户活动当天的平均时间或类似的事情-根据你的陈述 I need to know in which time is where the user do some actions in my app ```

standardSQL

SELECT TIMESTAMP_ADD(ts, INTERVAL avg_value MILLISECOND)
FROM (
SELECT TIMESTAMP_TRUNC(ts, DAY) ts, CAST(AVG(TIMESTAMP_DIFF(ts, TIMESTAMP_TRUNC(ts, DAY), MILLISECOND)) AS INT64) avg_value
FROM project.dataset.table
GROUP BY TIMESTAMP_TRUNC(ts, DAY)
)

您可以使用下面的示例中的虚拟数据来测试、播放上面的内容

standardSQL

WITH project.dataset.table AS (
SELECT TIMESTAMP '2020-07-13 00:01:12.616 UTC' ts UNION ALL
SELECT '2020-07-13 00:01:14.616 UTC' UNION ALL
SELECT '2020-07-13 00:01:15.616 UTC'
)
SELECT TIMESTAMP_ADD(ts, INTERVAL avg_value MILLISECOND)
FROM (
SELECT TIMESTAMP_TRUNC(ts, DAY) ts, CAST(AVG(TIMESTAMP_DIFF(ts, TIMESTAMP_TRUNC(ts, DAY), MILLISECOND)) AS INT64) avg_value
FROM project.dataset.table
GROUP BY TIMESTAMP_TRUNC(ts, DAY)
)

有输出

Row f0_
1 2020-07-13 00:01:14.283 UTC