Some of the required rigidity could be provided by the rotation alone, by what's the simplest to describe as a centrifugal force (it really doesn't exist, it's merely a product of other forces), and it would be simplest to limit lateral wobble, if the thrust is applied perpendicular to the rotation vector from all its extremes simultaneously, i.e. your non-rigid wheel rotates 90° to your velocity vector.
My first edit was wrong. Rotating in the axis along your velocity vector would not only complicate engine design, but also eventually collapse the structure as one part of the wheel decelerates while the other accelerates during one single rotation. So unless you could compensate for the propulsive force with magnetic fields to keep the extreme parts of the wheel at a stable distance, this wouldn't work. But rotating the wheel perpendicular to the velocity vector could. If the two forces (centrifugal from our rotation and propulsive from our engines) are kept at exactly the 90° angle, they're non-interactive and the extreme ends of the wheel would keep exactly the same distance to each other as if the engines were off. That's true because the centrifugal force vector is equal in all directions along the wheel's plane of rotation, and the force on each of its parts is canceled out by its opposing end.
Problem is, that it would be imperative to keep the thrust vector (your normal) perpendicular to the wheel's rotational plane, the two (or more, but that's even harder to do then) engines would have to be perfectly synchronised, and run in a mutually compensating mode, where all the other engines would have to adjust to the worst of them. So while the engines perform exactly as expected, this could be relatively simple. Simply ignore that the whole structure rotates too and consider all its extreme parts with engines as individual spaceships flying in precision formation. If your velocity vector is exactly perpendicular to the plane of rotation, these forces don't have any influence on each other As one engine starts acting up though, all the rest of them have to follow suit and compensate, otherwise you end up with and out-of-plane rotation, precession of its extreme parts and eventual collision between them.
So, in my opinion; doable, but complicated. A rigid structure with a central engine would be simpler...