Good question. I recommend that you test your thermoplastic material at three different strain-rates, for example 10/s, 1.0/s, and 0.1/s. These rates can often be reached in traditional testing machines. You can then use the results from these tests to extrapolate the response at higher rates using an appropriate material model. Note that there are very specialized test setups (e.g. the Split Hopkinson Bar test) that reach strain rates as high as 5000/s, but these tests are difficult to perform and expensive.
After you have experimentally tested your material you need to select an appropriate material model. The easiest model to use is likely an elastic-plastic model. There are, however, many material models that are significantly more accurate for thermoplastics. One example of these models is the Hybrid Model (HM). These powerful models are not typically built-in into the FE codes, but are available through external subroutines.
Best of luck,
Jorgen Bergstrom, Ph.D.