2

我正在使用 SQL Server 2008，如何计算开启和关闭状态之间的时间？我有下表（按顺序排列timestamp）：

ID | EQUIP_ID | TIMESTAMP           | STATUS (1 on/0 off)
1  |     1    | 21/05/2012 13:00:00 |   1
3  |     1    | 21/05/2012 13:04:00 |   1
4  |     1    | 21/05/2012 13:05:00 |   0
6  |     1    | 21/05/2012 13:09:00 |   1
7  |     1    | 21/05/2012 13:10:00 |   1
9  |     1    | 21/05/2012 13:12:00 |   1
10 |     1    | 21/05/2012 13:13:00 |   0
10 |     1    | 21/05/2012 13:14:00 |   1
10 |     1    | 21/05/2012 13:15:00 |   0

我期待这样的结果：

EQUIP_ID |    START             |          END           | STATUS
    1    | 21/05/2012 13:00:00  |   21/05/2012 13:05:00  |   1       (WORKING)
    1    | 21/05/2012 13:05:00  |   21/05/2012 13:09:00  |   0       (STOPPED)
    1    | 21/05/2012 13:09:00  |   21/05/2012 13:13:00  |   1
    1    | 21/05/2012 13:13:00  |   21/05/2012 13:14:00  |   0
    1    | 21/05/2012 13:14:00  |   21/05/2012 13:15:00  |   1

我已经尝试了一些间隙和岛屿的功能，但没有奏效，我不知道我错过了什么......

4

2 回答 2

2

这种方法首先过滤掉重复的行和重复的行。在只剩下状态开关的情况下，可以通过基于 搜索下一行来检索“结束时间” row_number。

; with  numbered as
        (
        select  row_number() over (partition by equip_id order by timestamp) rn
        ,       *
        from    YourTable
        )
,       nodups as
        (
        select  row_number() over (partition by cur.equip_id order by cur.timestamp) rn
        ,       cur.equip_id
        ,       cur.timestamp
        ,       cur.status
        from    numbered cur
        left join    
                numbered prev
        on      cur.rn = prev.rn + 1
                and cur.status = prev.status
        where   prev.id is null
        )
select  cur.rn
,       cur.equip_id
,       cur.timestamp as StartTime
,       next.timestamp as EndTime
,       cur.status
from    nodups cur
left join    
        nodups next
on      next.rn = cur.rn + 1
        and next.equip_id = cur.equip_id

SQL Fiddle 上的实时示例。

于 2012-07-27T19:03:54.433 回答
0

这是我的看法。假设您的表名为“MyData”：

WITH operating AS
(
    SELECT
        d.EQUIP_ID
        , d.[TIMESTAMP]
        , d.[STATUS]
        , ROW_NUMBER() OVER (PARTITION BY EQUIP_ID ORDER BY [TIMESTAMP]) RowNum
        , ROW_NUMBER() OVER (PARTITION BY EQUIP_ID ORDER BY [TIMESTAMP]) -
            ROW_NUMBER() OVER (PARTITION BY EQUIP_ID, [STATUS] ORDER BY [TIMESTAMP]) AS [Group]
    FROM 
        MyData d
)
SELECT 
    state1.EQUIP_ID
    , MIN(state1.[TIMESTAMP]) [START]
    , MAX(state2.[TIMESTAMP]) [END]
    , state1.STATUS 
FROM 
    operating state1
LEFT JOIN
    operating state2 
    ON 
    state1.RowNum = state2.RowNum - 1
WHERE
    state2.[TIMESTAMP] IS NOT NULL
GROUP BY  
    state1.EQUIP_ID, state1.[STATUS], state1.[Group]
ORDER BY 
    MIN(state1.[TIMESTAMP])

它利用该ROW_NUMBER()函数来确定每个EQUIP_ID. 然后它只是查找状态何时开始（MIN([TIMESTAMP])），然后我将其与下一行中的结束时间（MAX([TIMESTAMP])）匹配（请参阅自加入RowNum）。WHERE消除没有结束时间的最后一行。我得到的结果是：

EQUIP_ID | START                   | END                     | STATUS 
---------+-------------------------+-------------------------+-------
       1 | 2012-05-21 13:00:00.000 | 2012-05-21 13:05:00.000 |      1
       1 | 2012-05-21 13:05:00.000 | 2012-05-21 13:09:00.000 |      0
       1 | 2012-05-21 13:09:00.000 | 2012-05-21 13:13:00.000 |      1
       1 | 2012-05-21 13:13:00.000 | 2012-05-21 13:14:00.000 |      0
       1 | 2012-05-21 13:14:00.000 | 2012-05-21 13:15:00.000 |      1
于 2012-07-27T19:47:11.433 回答