您应该使用count（distinct books\u tag.tag\u id）

SELECT books_tag.book_id, books_pre.title
FROM books_tag
JOIN books_pre ON books_tag.book_id = books_pre.id
JOIN tags ON books_tag.tag_id = tags.id
WHERE tags.name IN ('a', 'd')
GROUP BY books_tag.book_id
HAVING COUNT(distinct books_tag.tag_id) = 2

尝试以下简短查询：
样本数据：

create table Books(id int, title varchar(10));
insert into Books values
(1, 'maths'),
(2, 'science'),
(3, 'HP');
create table Tag(id int, name char(1));
insert into Tag values                            
(1, 'a'),
(2, 'b'),
(3, 'c');
create table Book_tag(id int, book_id int, tag_id int);
insert into Book_tag values
(1, 1, 2),
(2, 1, 3),
(3, 2, 1),
(4, 3, 1),
(5, 3, 2);

t-sql：

select distinct book_id from Book_tag bt
where exists(select 1 from Tag
             where id = bt.tag_id
                   and name in ('c', 'a'));

部分 name in (...) 指定要搜索的标记。

您尝试的查询是 AND 标准像得到那些有标签a和标签d的书，根据你的问题你不需要这个标准，只要删除groupby和having子句，你就可以了

SELECT DISTINCT b.*
FROM books_pre b
JOIN books_tag bt ON bt.book_id = b.id
JOIN tags t ON bt.tag_id = t.id
WHERE t.name IN ('a', 'b')

演示