Machine learning methods for optimizing underwater separation of a carrier hood

This study establishes a finite element model to investigate the underwater separation of a carrier hood, focusing on the impact of the side-thrust rocket engine position and thrust direction on the separation time and attitude of the hood. It was revealed that unreasonable conditions can cause delayed separation or collisions with the launch barrel, thereby compromising the safety and reliability of the launch. To optimize this process, 91 working conditions were simulated and processed using a random forest machine learning model. The optimal condition is achieved when the horizontal coordinate of the side-thrust rocket engine's fixed position is 0.275 m, and the thrust direction is 15°, with the time required for the hood to leave the launch channel being 0.177 s. The error between the simulation results and the model predictions is only 1.14%, indicating the significant application value of this optimization method in submarine-launched missile underwater launches.