Hi,
I'd suggest to translate this accel impulse into a force impulse, then set up a transient analysis with timestep-variable force magnitude. If you choose enough timesteps within the impulse duration, the force variation will look like a sine kind nicely. If you know APDL, you can define a function to analytically define the force (it is a bit involved with Workbench... I've been told that this kind of input will be possible with ANSYS v.11)
Other way: build your FE model in Workbench, then exit to Classical and perform the analysis there (you have many more possibilities in this environment).
Regards