There are variations of Model Predictive Control that allow you to incorporate measured or predicted disturbances. I don't have any direct experience in this method, but I've been reading up on it recently. Worth looking at.
In the case of low frequency disturbances at the plant input ( for instance, wind acting over an antenna) , a simple solution is to add an integrator to the feed-forward controller provided that the plant has no such an integrator. If the disturbance is at the output ( for instance, measurement noise) then the noise effect can be cancelled or damped by allocating appropriate zeros- at the noise frequency or closed to it in average- as zeros of the output- feedback controller .
Hi there, what do you mean by 'known disturbance profile'? Do you mean that you know the exact time-series of the disturbance (equivalent of a reference signal), or are you talking about the power spectrum, or perhaps something else? Depending on what you know about your disturbance, you might be able to use other control laws than feed-forward. Typically these are based on the internal model principle: http://link.springer.com/article/10.1007%2FBF01447855
if it is possible to model the process and disturbance then a great choice will be some variation model predictive control MPC eg DMC or GPC. Check the MPC toolbox of MatLab.
As has been pointed out previously, you could just use a PID controller for this. If you know the power-density spectrum of your disturbance input, then when you tune your PID controller, you would need to make sure that the transfer function for your plant x controller product F(s), has a bandwidth that approximately matches that of the disturbance (in addition to having healthy stability margins). If F(s) has negligible gain at the dominant disturbance frequency, then your controller will be helpless and unable to reject it!
If you can model the disturbance behavior (as you mentioned the known disturbance profile) or even know the disturbance come into the view as a known signal, then you can use its model to observe or predict the disturbance and then combine this information with your control objectives (tracking or regulation, etc.) in an optimization fashion to obtain optimal control input. An excellent approach to do this is Model Predictive Control method.