There are different models of uncertain data in literature, and they are not compatible data types.
UKMeans needs discrete samples from uncertain data, it does not process continuous distributions directly. You need to use a DiscreteUncertainifier to produce such discrete samples from the uncertain data.
See the JavaDoc of the class hierarchies and filters:
https://elki-project.github.io/releases/current/doc/elki/data/uncertain/package-summary.html