Abstract
Abstract This paper proposes a novel global sensitivity analysis method that combines deep neural networks with saltelli estimation, significantly enhancing the efficiency and reliability of sensitivity analysis in complex mechanical systems. Given that sensitivity analysis often requires a large amount of sample data, and the high-precision multi-body dynamics simulation model of tanks inevitably leads to increased computational costs. To address this, based on electromechanical integrated dynamics model of tank stabilization system, a deep neural network surrogate model for muzzle stability accuracy was established. Subsequently, further integrating saltelli estimation, a global sensitivity analysis targeting muzzle stabilization accuracy was conducted by selecting 18 key parameters of the actuator electric cylinder as the object. Finally, a key parameter system of the actuator electric cylinder influencing muzzle stabilization accuracy was established, providing theoretical support for the design optimization of the next generation tank stabilization system.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call