what is it that leads you to believe that a '51 derivative could not take 4 inputs and generate some outputs for actuators?
If the app is as simple as what you "distill" it to, the I have totally misread the OP. The fuzzy part of both the programming and the description makes me doubt.

While I can, indeed see a final model controlled by a (collection of) '51 I doubt that with the lose parametres specified so far that an initial devlopment can be done on a '51.

Had the specification been complete tednically (such as e.g. for a series of pulses, based on the frequency an actuator shall assume a corresponding angle) I might have stated otherwise, but it seems that the first unit will not be a "working unit" but a "figuring out unit" and for that I would prefer something with a bit of horaspower since, who knows, for the initial investigation it might be that the initial idea is that a floting point fuzzy decision must be taken in 1ms. If, after working everything out, it is shown that the requirements can be fulfilled by cheaper means, then a (few) '51 may very well be right.

As to using a '51 I even suggested the possibility of "a network of small, cheap processors" and, there, I should have clarified that I ment '51s.

Erik