PHOTOVOLTAIC SYSTEMS (PV) ARE BY NATURE NONLINEAR SOURCES. THEIR POWER OUTPUT VARIES DEPENDING MAINLY ON THE LEVEL OF SOLAR RADIATION AND AMBIENT TEMPERATURE. THE MAIN DRAWBACKS OF THE PV SYSTEMS ARE HIGH FABRICATIONS COST AND LOW ENERGY-CONVERSION EFFICIENCY, SINCE THEY RARELY WORK IN MAXIMUM POWER POINT (MPP).VARIOUS METHODS HAVE BEEN PROPOSED AND IMPLEMENTED TO REALIZE MPP. SOME OF THEM ARE: NONLINEAR EQUATIONS AND MATHEMATICAL MODELING. THESE APPROACHES HOWEVER REQUIRE DETAILED KNOWLEDGE OF PHYSICAL PARAMETERS RELATING TO THE SOLAR-CELL MATERIAL AND MANUFACTURING SPECIFICATIONS. AS SUCH INFORMATION MAY NOT BE READILY AVAILABLE TO THE USERS OR, HAS GIVEN FOR A SMALL RANGE OF V-I CHARACTERISTIC.IN THIS THESIS A NEURAL NETWORK FOR SOLAR-CELL MODELING AND MPPT IS MULTILAYER PERCEPTRON NETWORKS PROPOSE. THE NUMBER OF LAYERS FOR NETWORKS IS 3: THE INPUT LAYER, THE HIDDEN LAYER AND THE OUTPUT LAYER. THE NETWORK INPUTS FOR MODELING ARE: RADIATION, TEMPERATURE AND LOAD VOLTAGE; AND THE NETWORK OUTPUT IS CELL CURRENT. THE NETWORK INPUTS FOR MPP PREDICTION ARE: RADIATION AND TEMPERATURE; AND THE NETWORK OUTPUT ARE VOLTAGE AND CURRENT CORRESPONDING TO MPP.AFTER IMPLEMENTATION NETWORK, THE DATA COLLECTED FROM A REAL SOLAR-CELL MANUFACTURED BY THE IRANIAN OPTICAL FIBER FABRICATION CO. IS USED FOR TRAINING AND TEST THE NETWORK. THE RESULTS OF NETWORK WILL EXPLAIN.