This paper proposes a power amplifier behavior modeling based on neural network. This power amplifier uses genetic algorithm to find the suitable weights and number of neurons. In this paper a real valued time delay neural network has been trained and optimized with hybrid genetic algorithm. This neural network is used as a pre distorter and model a radio frequency power amplifier. The simulations show that the training is done with higher speed and sufficient linearity. Result shows that hybrid genetic algorithm can do with more speed than back propagation. The performance of the amplifier is optimized for input signals with frequencies in a range of radio frequency. The neural network is used as pre distorter of power amplifier and showed 9 db improvement in linearity in response to a wide band signal.
