Abstract: Muzzle velocity degradation prediction of machine gun is a complicated non-linear problem. Generalized regression neural network (GRNN) has been widely used in the modeling of the non-linear problems, but GRNN has rarely been used to predict the muzzle velocity degradation of machine gun. Since the smoothing factor of GRNN obviously affects the prediction performance of neural network, the fruit fly optimization algorithm is used to automatically select the parameters of GRNN. A method to model a muzzle velocity degradation based on general regression neural network with fruit fly optimization algorithm (FOAGRNN) is proposed. A prediction model is established based on the experimental data of muzzle velocity degradation, in which the muzzle velocity degradation is taken as characteristic quantity. The predicted results are basically consistent with the experimental results. The research result shows that FOAGRNN model outperforms GRNN model with default parameter and BPNN prediction model in the prediction of muzzle velocity degradation. Key

Key words: ordnancescienceandtechnology, fruitflyoptimizationalgorithm, generallizedregressionneuralnetwork, muzzlevelocitydegradation, predictionmodel