有限元模拟结合CPO算法的晶圆搬运机器人钢带疲劳寿命预测

Fatigue Life Prediction of Steel Belts in Wafer Handling Robots Using Finite Element Simulation Combined with CPO Algorithm

摘要: 晶圆搬运机器人机械臂中的牵引式钢带在有限角度内工作，有限带长受到的循环交变应力作用十分频繁，极易发生疲劳破坏，目前针对此类特殊传动结构的疲劳寿命预测研究较少。针对这一问题，本研究建立了一种将有限元分析和神经网络预测技术相结合的预测方法。首先使用Workbench有限元软件确定钢带结构的薄弱位置，基于S-N曲线修正理论和Basquin方程建立钢带疲劳寿命数学模型。通过nCode DesignLife软件计算钢带的疲劳寿命，以确定钢带寿命较短的区域，最后采用CPO算法对传统BP神经网络进行优化并预测钢带的疲劳寿命。结果显示，经过CPO算法优化的神经网络预测结果与nCode分析误差不超过5%，误差远低于传统BP神经网络。该研究对晶圆搬运机器人钢带的疲劳寿命的预测具有一定的意义。

Abstract: In wafer handling robot manipulators, the traction steel belt operates within a limited angle and is subjected to frequent cyclic alternating stresses due to its constrained length, making it highly susceptible to fatigue damage. Currently, there is limited research on fatigue life prediction for such specialized transmission structures. To address this issue, this study proposes a predictive approach that integrates finite element analysis with neural network-based prediction techniques. First, Workbench finite element software is used to identify vulnerable locations in the steel belt structure. A mathematical model for the fatigue life of the steel belt is established based on the modified S-N curve theory and the Basquin equation. The fatigue life of the steel belt is then calculated using nCode DesignLife software to identify regions with shorter service life. Finally, the CPO algorithm is employed to optimize the traditional BP neural network for predicting the fatigue life of the steel belt. The results show that the prediction error of the neural network optimized with the CPO algorithm is within 5% compared to the nCode analysis, which is significantly lower than that of the traditional BP neural network. This research provides valuable insights into predicting the fatigue life of steel belts in wafer handling robots.