Whatever optics design you use, it can only intercept a certain amount of solar energy, and that determines its effectiveness. If the lens is very large, it might sound like a good idea, but optically it isn't very useful. Applied to a whole panel it will cause the irradiance on the cells to vary widely and you lose effectiveness when this happens. In the other extreme it is equivalent to covering the panel with very large numbers of small hemispheres; but that doesn't work because the total area of the array seen from a distance hasn't changed much. With panels costing what they do, it would be cheaper to use two panels rather than one panel and an optical device like a lens.
The one area where lenses (or preferably mirrors) are essential is in concentrated PV. But such systems have other issues, like directional sensitivity, spectral sensitivity, overheating and so on.
Khalid: I think you may benefit more by studying the split between direct irradiance and reflected irradiance. You may be surprised by the quantum of irradiance that reaches a PV collector post reflection from clouds and surroundings.