Is it possible to merge EPIC and 27k methylation array data and carry out a normalization of the combined data? I merged the two data using combineArrays function from MINFI. But I had to convert the arrays from RGChannel set to methyl set in order to do this. Now I cannot normalize the data in methyl set form using preprocessFunnorm as it needs an RGChannel set as input.

Combining and then normalizing is almost surely a bad idea. A slightly better idea would be to normalize separately and then try to combine, and likely better still would be to keep the data separate up to the point where you make comparisons and then look for DMRs that are consistent between the two groups.

This assumes that you are making comparisons within each group and not trying to make comparisons between the EPIC and 27k data. If that's what you are trying to do, you should stop doing it because that's a fool's errand.