In this paper, a scheme for estimating the parameters of solar cells and a NN-based emulator for predicting the maximum power point are presented. The diode model with series and shunt resistors is used to estimate parameters highly affecting its V-I characteristic curve and both a real-coded genetic algorithm and the model adjustment technique are employed. For implementing the emulator, a multi-layered neural network incorporating with the BP algorithm is used. A set of simulation works using both field data and generated data are carried out to demonstrate the effectiveness of the proposed method.