Surface plasmons are not strictly just electromagnetic waves. They can be simplistically visualized as coherent oscillations of electrons at the interface. Some spectrum of incident light can launch surface plasmons at some of the solar cell's interfaces. The simplest mechanism will be energy transfer from plasmon to form an exciton - and then the story is the same as with light.
Another issue is field enhancement - which can be induced by plasmons. Higher fields can lead to higher rates of exciton formation.
The mechanism seems to be similar to as SERS and SEF, which says that in the vicinity of the molecule, the field gets enhanced by the plasmonic structure. This enhanced field leads to enhanced absorption.
The extent of enhancement of solar cells with current plasmonic structures, though is not much promising, because of their small lifetime.
In standard semiconductor solar cells, next to field enhancement or PRET (plasmon resonant energy transfer resulting in the creation of excitons), simple scattering effects are exploited for either (i) increasing the optical path lenght in particular for the NIR for metal nanoparticles (MNPs) placed on the front side or (ii) backscattering effects for MNPs at the rear side of the cell.
Other types of solar cells, such as synthetic dyes or organic polymer cells, can equally benefit from plasmon assisted field effects [see for instance APL 93, 073307 (2008)]. The advantage is, that the MNPs can be placed directly inside the dye or polymer, with electro- or chemical deposition. Then field enhancement effects can be exploited (i.e. higher fields, larger energy transfer, but unfortunately also just a lot of ohmic heat produced) which is more tedious to achieve in standard Si cells. Interestingly, this is compatible to plasmon assisted enhancement of photoluminescence which can also yield a downshifting of contributions from the NIR.