The concepts of gain and phase margins are only defined for linear time invariant (LTI) systems and controllers. Since fuzzy controllers are nonlinear controllers, stability margins can not be defined for such controllers (yet!). Usually, the performance and robustness of the system under the designed controller is measured from the time-domain response of the system. I've seen couple of papers proposing model based fuzzy controller (MBFC). It is a linear fuzzy based controller in which the stability margins can be evaluated, but the approach doesn't apply to regular fuzzy controllers. That's the main criticism from control community on fuzzy controllers.
if your system is not a complex time invariant one, you can approximate it by an analyzable system. e.g. you can get the step/impulse response of your system and apply it in 'ident" toolbox of matlab. after finding a similar model, analyze it by bode/nyquist plot.