Comparing the root-mean-square of deviation (RMSD) between the simulated and crystallised structure in Protein Data Bank will be a good start. You can use PyMOL for the purpose.
In order to compare your predicted structure with an actual structure what you would have to do would be to have the structure of the protein already solved in the protein data bank.
To experimentally determine the structure the main tools are very pure protein + X-ray crystallography or cryo-electron microscopy or Nuclear magnetic resonance and a lot of math.
What you could do is take 5 structures in the protein data bank.
Try to fold the protein from the fasta files. Compare your algorithms with the actual structure using the procedure Zulkeflee Sabri describes.
But if you try to fold the linear protein with the PDB file of the experimental structure then that is cheating.