Looks like a harmonic function, probably some kind of step-like instead of a sinewave; there's drift in the amplitude and frequency over time (looks like amplitude is increasing and frequency is decreasing), as well as a trend downards that may be linear but I guess might be logrithmic. Well, it might have some kind of second or third order behavior as a sinusoid rather than actually being a modified square wave.

Anyway, I doubt it's exactly a polynomial.

I feel like figuring out what kind of equation is governing this behavior is too fuzzy for a computer. Once you've got the formulation you can drive the coefficients down with least squares + solver, but the human needs to figure out what's going on here.