在使用MySQL進行開發和管理時，我們經常需要查看數據庫中耗費最多CPU的語句，以便優化和改善數據庫性能。在MySQL中，我們可以通過以下語句來查詢：

SELECT 
proc.PROCESSLIST_ID AS "線程ID",
proc.USER AS "用戶",
proc.DB AS "數據庫名",
proc.COMMAND AS "命令",
proc.TIME AS "執行時間",
proc.STATE AS "狀態",
CONCAT(proc.INFO,";",sql_text) AS "執行語句" ,
ROUND(proc.CPU_TIME,6) AS "CPU時間（秒）"
FROM performance_schema.threads AS threads
LEFT JOIN performance_schema.events_statements_current AS statements ON statements.THREAD_ID=threads.THREAD_ID
LEFT JOIN (
SELECT ID,CONCAT('"',REPLACE(SUBSTRING_INDEX(info,' ',3),'"',''),'"') AS INFO FROM performance_schema.events_statements_history_long
) AS info ON info.ID=IFNULL(statements.EVENT_ID,-1)
LEFT JOIN (
SELECT PROCESSLIST_ID,EXECUTION_TYPE,STATE,INFO as MAX_INFO
FROM performance_schema.threads AS threads
INNER JOIN performance_schema.events_statements_history_long AS statements ON threads.PROCESSLIST_ID=statements.THREAD_ID 
INNER JOIN (
SELECT PROCESSLIST_ID,MAX(SUBSTRING_INDEX(INFO,'during:',-1)) AS MAX_DURATION 
FROM performance_schema.threads AS threads
INNER JOIN performance_schema.events_statements_history_long AS statements ON threads.PROCESSLIST_ID=statements.THREAD_ID 
WHERE TIME>CURRENT_TIMESTAMP-INTERVAL 60 SECOND AND SQL_TEXT IS NOT NULL 
GROUP BY PROCESSLIST_ID
) AS DURATIONS ON DURATIONS.PROCESSLIST_ID=PROCESSLIST_ID AND MAX_INFO LIKE CONCAT('%',MAX_DURATION,'%')
) AS proc ON proc.PROCESSLIST_ID=threads.PROCESSLIST_ID
WHERE threads.PROCESSLIST_ID IS NOT NULL 
ORDER BY proc.CPU_TIME DESC
LIMIT 10;

以上語句使用了performance_schema系統庫中的事件表，聯合查詢了threads、events_statements_current、events_statements_history_long表，并通過CPU_TIME字段進行排序，列出SQL語句CPU消耗最高的10條語句。

在實際使用中，我們可以結合具體情況進行優化和改善，例如針對性地對查詢語句進行索引優化、減少冗余數據等。