Abstract:Addressing the problem of high reliability and limited failure data of firearms, a method for predicting the remaining life of gun barrels is proposed by using the initial velocity of the projectile as a characterization of the degradation of the barrel. We establish a degradation model for the initial projectile velocity based on a nonlinear Wiener process and create a gun barrel residual lifespan prediction model that takes into consideration the randomness of the life threshold. We employ the maximum likelihood method and Bayesian method for the estimation of degradation parameters and random failure threshold parameters. We conduct an analysis using a 14.5mm anti-aircraft machine gun as our subject. We obtain the estimated initial projectile velocities for a 14.5mm anti-aircraft machine gun under different projectile counts. We compare and analyze the prediction results based on the linear Wiener process prediction method, the nonlinear Wiener process prediction method, and the nonlinear Wiener process with random threshold prediction method. The results indicate that the accuracy of this method is higher than the other two methods, which can provide reference for improving the reliability of firearms in the context of big data.