What is the convention in Stack Exchange with respect to accepting questions that come with answers by the same author?
If you post a question, and nobody else gives a sufficient answer and you've found the answer on your own, then you are strongly encouraged to post that answer and accept it.
If you're posting a question for the sole purpose of answering it to share some information, then that's encouraged by the greater SE network too, but remember, if you don't like the way someone's doing that, you're allowed to vote in almost any way you please.
There is one significant difference, and that is that self accepted answers are not automatically pinned at the top.
If there is a higher voted answer, then it will appear first.
I assume you mean "accepting" as in "giving a positive or negative response", as opposed to "clicking the green tick" (marking as accepted). Apologies if this isn't the case.
If the question is genuine, and the answer involved putting together information in a useful way, then I am happy at people doing so. This is an example of a question that I, in my own opinion, fits this scenario: Why are there so many unscheduled/premature elections in Israel?
However, if the person already knows the answer to the question, and their answer merely consists of extra publicity to a piece of information that is already readily accessible, then I am less likely to upvote them, and more ready to downvote or closevote them so long as there's other factors that would justify a downvote/closevote. As said by a diamond moderator on a self-answered question on Skeptics.SE:
My guess (and the downvotes aren't mine, so it is a guess) is that it is a reaction to the Evan Carroll's habit of posting politically-motivated questions and then immediately posting prepared answers. It gives the strong impression that the site is being treated as a personal blog, rather than for genuinely finding answers to open questions. At this stage, there is no rule against such behaviour.
But that's just how I vote - it's not a rule or anything.