Classical utilitarians usually argue for their view from an impartial, altruistic perspective. If they had atypical intuitions about particular cases, they would discard them if it can be shown that the intuitions don't correspond to what is effectively best for the interests of all sentient beings. So in order to qualify as genuinely other-regarding/altruistic, the procedure one uses for coming up with a suffering/happiness exchange rate would have to produce the same output for all persons that apply it correctly, otherwise it would not be a procedure for an objective exchange rate.

The procedure you propose leads to different people giving answers that differ in orders of magnitude. If I would accept ten hours of torture for a week of vacation on the beach, and someone else would only accept ten seconds of torture for the same thing, then either of us will have a hard time justifying to force such trades onto other sentient beings for the greater good. It goes both ways of course, if classical utilitarianism is correct, too low an exchange rate would be just as bad as one that is too high (by the same margin).

Since human intuitions differ so much on the subject, one would have to either (a) establish that most people are biased and that there is in fact an exchange rate that everyone would agree on, if they were rational and knew enough, or (b) find some other way to find an objective exchange rate plus a good enough justification for why it should be relevant. I'm very skeptical concerning the feasibility of this.

Preference utilitarianism is not the same thing as hedonistic utilitarianism (they reach different conclusions), so you can't use one to define the other.

How do you measure the intensity of desires, if not by introspectively comparing them (see below)? If you do measure something objectively, on what grounds do you justify its ethical relevance? I mean, we could measure all kinds of things, such as the amount of neurotransmitters released or activity of involved brain regions and so on, but just because there are parameters that turn out to be comparable for both pleasure and pain doesn't meant that they automatically constitute whatever we care about.

If you instead take an approach analogous to revealed preferences (what introspective comparison of hedonic valence seems to come down to), you have to look at decision-situations where people make conscious welfare-tradeoffs. And merely being able to visualize pleasure doesn't necessarily provoke the same reaction in all beings -- it depends on contingencies of brain-wiring. We can imagine beings that are only very slightly moved by the prospect of intense, long pleasures and that wouldn't undergo small amounts of suffering to get there.