Yes, IMHO accelerometers are a better way to go initially.

This sort of thing has been done in commercial products, years ago
although I don't know the details.


Stangely enough, this is what I do for a living. No kidding.
I am an engineer at Bently Nevada, Corp, the industry leader in machinery
diagnostics.
Most of our claim to fame is for rotating machinery; however, we do quite a
bit of work with reciprocating engines as well.

An accelerometer will work well.
So will proximity (vibration) probes.
We do all kinds of little tricks like putting a notch in the crankshaft to
use as a phase reference.
Then we can determine at what point in the cycle an anomaly occurs, which
can lead to different diagnoses.
We do a whole lot more, but I can't give away all of the trade secrets ;-)

Adam
N7966L
Beech Super III