例如，您应该能够使用条件聚合来实现这一点

drop table if exists t;

create table t(item int,qty_out int , dt date);

insert into t values
(1,1,'2018-09-01'),(1,1,'2018-10-01'),(1,1,'2018-11-01');

select item,
         sum(case when year(dt)*12 + month(dt) = year(now()) * 12 + month(now()) then qty_out else 0 end) thismm,
         sum(case when year(dt)*12 + month(dt) = (year(now()) * 12 + month(now()) -1) or
                            year(dt)*12 + month(dt) = (year(now()) * 12 + month(now()) -2) then qty_out else 0 end) last2mm,
         (sum(case when year(dt)*12 + month(dt) = year(now()) * 12 + month(now()) then qty_out else 0 end) +
         sum(case when year(dt)*12 + month(dt) = (year(now()) * 12 + month(now()) -1) or
                            year(dt)*12 + month(dt) = (year(now()) * 12 + month(now()) -2) then qty_out else 0 end)
         ) / 3  amc
from t
where year(dt)*12 + month(dt) >= (year(now()) * 12 + month(now()) -2)
group by item ;
+------+-----------+-------+--------+
| item | thismonth | last2 | amc    |
+------+-----------+-------+--------+
|    1 |         1 |     2 | 1.0000 |
+------+-----------+-------+--------+
1 row in set (0.01 sec)

请注意转换为月数以简化日期在年终的位置。当然，如果你正在寻找一个3个月的滚动平均值，那么这将是一个不同的问题。