Optimization of small caliber projectile based on neural network

Abstract

When small caliber projectile is moving at high speed underwater, the water around the projectile will cavitate. The geometric shape of the warhead with the best drag coefficient corresponds to the supercavitating state where the projectile is completely enveloped by cavitation. In this paper, aiming at a small-caliber projectile, the three-section cone is selected as the basic projectile, and the shape of the projectile is optimized with the drag coefficient as the optimization objective. The neural network and sequential quadratic programming (SQP) algorithm are combined to reduce the calculation amount in the optimization process and improve the optimization efficiency. The drag coefficient of the optimized projectile is reduced by about 40% compared with the projectile before optimization, and it can form a supercavitation that envelops the entire projectile.