I would not abandon the idea of giving a talk about your own research or the circle of ideas that motivate it. Almost everything in pure mathematics is ultimately motivated by questions that a broad audience might appreciate. What are the patterns in the distribution of primes? Is every curve an intersection of two surfaces? Which numbers are sums of two squares? How many fundamentally different ways are there to continuously attach a line to each point on a circle? What is the smallest area in the plane inside of which it's possible to continuously turn a unit line segment around by 180 degrees? Almost surely your research arises from some such natural question and its natural generalizations. I'd try to think of a good, simple, motivating problem, explain why it's a natural thing to wonder about, and where one might get stuck, and what kinds of new ideas this requires, and how one is naturally led to generalize, etc.
You've got the opportunity to help people appreciate, in some small way, what you actually do. Why waste it?
