The Underwater Projectile Launching Process Prediction Based on Reduced Order Method

In this paper, computational fluid dynamics (CFD) and reduced order method (ROM) are developed for predicting the velocity time series during the initial stage of projectile launching. The overlapping grid technique is firstly applied to perform high accuracy CFD simulations of the underwater launch of a submarine-launched projectile under two-parameter varying working conditions (i.e., different piston velocities and pipe interface radii). After that, the proper orthogonal decomposition (POD)-based ROM is used for decomposing the launching process into several spatial dependent mode functions and the corresponding time coefficients. The prediction of the initial stage of the launching velocity (which is the most important part of the launching process) is realized by establishing the neural network between the working condition parameters and the POD basis coefficients. The analysis results show that this method of CFD and ROM can accurately predict the complex dynamic process such as projectile launching process.