I think you are getting some things confused here. On Illumina arrays, most genes only have one probe associated with it. There are multiple replicates of a single probe (will share the same probe sequence). Occasionally, you will get two probes per gene (is this what you mean by replicate?). With Affy probe designs, you will get multiple probes per gene (probeset) but each probe might only be represented once (or very few times). Illumina's probe design (one probe, multiple copies) has been shown to be more accurate. If you want to learn more about this topic, I would have a look at the beadarray and lumi bioconductor manuals. I believe lumi was the one that introduced the concept of nuIDs but I have personally never found a use for them.

For your question, you will need to subset down to the few genes with multiple probes for that gene, typically if a gene has more than one probe it will be due to things like known alternative exon usage and alternative start sites. You can probably find all the info you need in this annotation set: http://www.bioconductor.org/packages/release/data/annotation/html/illuminaHumanv4.db.html