Is noumenon : phenomenon :: substance : accident ?
cf. "We know substances by means of their accidents?"
I realize the analogy isn't perfect: Aristotle thinks substances can be known, whereas Kant's noumena are unknowable Dingen an sich. But if Aristotle thinks substances are known by their accidents, why for Kant aren't noumena known from their corresponding phenomena?
Well, there's several things that break the analogy.
First, noumenal reality is not known at all for Kant. This is because knowledge for Kant has a very different meaning that it does for Aristotle. On Aristotle's model we know things as we perceive at them and reproduce a phantasm of them in our minds from which extract the essence and have knowledge in this way of the substance through its essence and particulars [i.e., accidents].
For Kant, we know things through the categories of understanding and the forms of sensibility which turn a thing into an object and phenomenon proper respectively (see the SEP entry for more details on the latter). Thus, for Kant things-in-themselves are not knoweable de minimis because they are not objects. They are inaccessible to the knowledge operation. In Aristotelian vocabulary, things in themselves are closer to absolute potential or prime matter. But they differ insofar as what comes after them is not exactly formed matter.
Second, while substances, for Aristotle, are formed matter and thus when we can know the form and the accidents of a substance, the relationship between thing and phenomenon and object is not one in which the thing has inherent properties that follow from its kind. As far as we can known on Kant's account, it makes no sense to speak of kinds of things for Kant. Because thingness doesn't admit kinds -- since kinds are categories.
Third, it's not entirely clear (here just inside of the Kantian picture) that we should identify noumenon with thing-in-itself. The noumenal is the realm of mind or reason -- which is not necessarily identical with the substrate of the phenomenon we perceive or the objects know. Or to put it another way, thing and object are a pair of opposites and phenomenal and noumenal are a pair of opposites as well, but it's not clear that what we do not know as objects is therefore noumenal. (This is a point on which there is some disagreement among Kant scholars).
One aside is that I tend to agree with what my advisor wrote many years ago in Kant Studien. God, on the Kantian picture, does have direct knowledge of things-in-themselves just as they. This is because God does not know through the categories of understanding.
I could go for that, but only partially.
Primary substance is real identity the way that a noumenal identity is an ultimately real thing that is the thing's reality, and the details of that primary substance are refracted into a number of secondary substance categories with the remainder captured as innumerable accidents, perhaps in the way that phenomenal reality may be shaped by noumenal qualities, with the difference made up of mere and absolute phenomena.
However the secondary substance categories are still substance (and may be more typical of a substance, as we think of one, than primary substance) whereas the emanations of noumenal reality into phenomenal reality are already phenomena. So the dividing point is not similar and the relationships are not quite parallel.
It is also merely an analogy and not a real similarity, the two distinctions do not divide the same world. Substance is real and attainable, not something mystical that we may or may not ever make contact with.