Do some normalization so that you can write queries more easily and won't have such anomalies.
You'll need another structure, like:
The element table
+----+---------------+
| id | name |
+----+---------------+
| 1 | some_element |
+----+---------------+
| 2 | another_elem |
+----+---------------+
| 3 | third_elem |
+----+---------------+
The part table
+----+----------+
| id | name |
+----+----------+
| 1 | wheel |
+----+----------+
| 2 | bearing |
+----+----------+
| 3 | hinge |
+----+----------+
| 4 | servo |
+----+----------+
etc..
And another, such as element_parts to connect the other two by an m:n relation
+----+---------+---------+
| id | elem_id | part_id |
+----+----------+--------+
| 1 | 1 | 1 |
+----+---------+---------+
| 2 | 1 | 2 |
+----+---------+---------+
| 3 | 1 | 3 |
+----+---------+---------+
| 4 | 2 | 3 |
+----+---------+---------+
| 5 | 2 | 4 |
+----+---------+---------+
etc..
And now you can write a query to, say, filter elements that contain (or need) wheel and servo (adapting this question's accepted answer):
select *
from element e
where 2 = (
select count(distinct p.id)
from element_parts ep
inner join part p on p.id = ep.part_id
where p.name in ('wheel', 'servo')
and ep.elem_id = e.id
);
