Have you tried fiddling with the timestep?
I like to experiment with simulations of switched coil converters. On powerup, boost converter output usually soars high for several cycles, then settles. There is little opportunity for periodic oscillations, due to a diode being inline with the inductor and capacitor.
A possible issue is that feedback may be 'loose', in the sense there is excessive delay to correct the duty cycle, after a change has occurred somewhere in the circuit. Could this be the case in your model?
- - - Updated - - -
Because if feedback is 'loose', there is a tendency to overcorrect. Then the feedback signal (which is delayed) instructs 'correct in opposite direction.' Etc. Oscillations can develop.