Polarimetry springs to mind for concentration of sugars, though for mixtures it would be ambiguous.
I don't know how easy it is to find basic reference data, but you could always do your own calibration for the sugars you're interested in. Methods and theory are widely discussed.
I haven't worked out why you want a "non-intrusive method", or really what you mean by that. It might help if you were a little clearer about what you are doing.
If you are making cups of coffee and don't want to rob the customer of even 1ml, then if your production method is standard, just divert one cup every so often to check that the amount of sugar you put in is actually reaching the cup. If you have a continuous flow production, then equally remove a sample occasionally: that won't intrude nor affect the quality of the main flow. The cost of testing is going to be several orders of magnitude more than the value of lost product.
Equally, if it's an automatic beverage machine, it's probably much better to improve the production process, rather than monitor varying quality outputs. "You can't test quality into a product".