Abstract: Solar Proton Events (SPE) are one kind of the major solar energy release phenomena. The released energies are well known from the magnetic energy. Lots of observations and studies have suggested that solar flares and CME (Coronal Mass Eruptions) are closely correlated to the complexity and non-potentiality of solar magnetic fields. Applying SOHO/MDI full-disk photospheric magnetograms, three magnetic physical properties (the maximum horizontal gradient of longitudinal magnetic field Bz, the length of neutral line with strong gradients L, the number of singular points η) are calculated, which describes the complexity and non-potentiality of solar magnetic fields. In order to verify that solar photospheric magnetic properties are available at forecasting SPE, a simple SPE forecast model that can forecast whether SPE will happen or not in the future 24 hours is set up, which is based on the three physical properties with BP (Back Propagation) neural network. In the consecutive tests for the 2002 and 2003 year samples, the model has very high accuracy rates (90 % for 2002 and 87.54 % for 2003, respectively) and the rather high SPE probabilities of detection (60 % for 2002 and 75 % for 2003, respectively), which supported that solar photospheric magnetic properties are very available in the SPE forecast models.