OPTIMAL DESIGN OF RUDDER STRUCTURES BASED ON DATA-DRIVEN METHOD

In the field of aerospace engineering, there is a growing demand for lightweight design based on structural topology optimization; specifically, the rudder structures of many high-tech equipment serve in severe thermal and mechanical environments, and it is both challenging and important to carry out efficient lightweight design of them. For a given load, the stiffness of thin-walled structures can be significantly enhanced by introducing stiffening ribs or reinforcing stiffeners, and this design philosophy is well consistent with the design requirement of air rudders. However, traditional stiffening ribs design under an implicit topology optimization framework suffers from a huge number of design variables, low computational efficiency, difficulty in guaranteeing the geometric characteristics of stiffening ribs, and inconvenience in directly importing optimization results into CAD systems. In this study, a new explicit topology optimization method (i.e., moving morphable component (MMC) method) is adopted, and combined with the data-driven methodology for efficient design of stiffening ribs in air rudders with irregular closed geometry. This method directly optimizes the geometric information of the stiffening ribs, and has the advantages of fewer design variables, high computational efficiency, and seamless connection between the optimization results and CAD software, so as to solve the issues of long optimization cycle and strong dependence on the experience of designers faced with the implicit topology optimization method and subsequent redundant steps such as model reconstruction and parameter optimization. Furthermore, the mapping between rib layout and key mechanical properties is described through an artificial neural network model, and used as a surrogate model for optimization to efficiently obtain high-quality initial designs, thus significantly improving the efficiency of the design optimization of rudder structures. The design framework presented integrates the structural topology optimization with artificial neural network model, which can be applied to the intelligent design of other key equipment.