A wheel shape optimization of a railway wheel cross section by means of Genetic Algorithms (GAs) is presented with the aim of minimizing rolling noise radiation. Two different approaches have been implemented with this purpose, one centred on direct Sound poWer Level (SWL) minimization, calculated using TWINS methodology, and another one emphasizing computational efficiency, focused on natural frequencies maximization. Numerical simulations are carried out with a Finite Element Method (FEM) model using general axisymmetric elements. The design space is defined by a geometric parametrization of the wheel cross section with four parameters: wheel radius, a web thickness factor, fillet radius and web offset. For all wheel candidates, a high-cycle fatigue analysis has been performed according to actual standards, in order to assure structural feasibility. Rolling noise reductions have been achieved, with a decrease of up to 5 dB(A) when considering the wheel component. Response surfaces have been also computed to study the dependency of the objective functions on the geometric parameters and to test the adequacy of the optimization algorithm applied.

提出了一种利用遗传算法（GAs）优化铁路车轮横截面的车轮形状的方法，目的是使滚动噪声辐射最小。为此目的，已经实现了两种不同的方法，一种以使用TWINS方法计算出的直接声功率级（SWL）最小化为中心，另一种以自然频率最大化为重点的强调计算效率。使用通用轴对称元素，通过有限元方法（FEM）模型进行了数值模拟。设计空间由具有四个参数的砂轮横截面的几何参数化定义：砂轮半径，腹板厚度系数，圆角半径和腹板偏移。对于所有候选车轮，已经根据实际标准进行了高周疲劳分析，为了确保结构上的可行性。滚动噪声已实现降低，当考虑车轮部件时，可降低高达5 dB（A）。还已经计算了响应面，以研究目标函数对几何参数的依赖性，并测试所应用优化算法的适当性。