Well, it's not that hard, once you set up the numerical problem with software.
The first thing you need is heliocentric ephemeris for both the asteroid and for the earth. Using this, you can find the asteroid ephemeris relative to the earth.
Then you need accurate star ephemeris (right ascensions and declinations vs. time) relative to earth.
Then you just step through the tables, and takes the angle between the radius vectors at each step. If the engle falls below (or gets really close) the angular radius of the asteroid, you have a possible occultation.
At that point you go back a few time steps, decrease the time step size, and go forward to get a more precise measurement.
To see if it will be visible from a particular location on the earth's surface, you calculate the ephemeris of the ground site relative to earth center and do the above again. This time, though, you need to see if the local elevation is positive, or the occultation will be below the horizon. Also, check to see if the local time will be at night, or else you won't see anything but blue sky and clouds.
Sounds complicated, but you could do it in Orbiter.
