Real-time Stress Prediction of Robotic Arms Based on Random Forest Agent Model

Digital twin technology plays a crucial role in the development of robotic arms, enabling accurate simulation and optimization of robotic arm performance by creating a virtual copy of the physical system [1]. In this paper, we innovatively combine finite element simulation with machine learning to explore a real-time prediction method for the stress of a robotic arm. The study first established a simplified 3D model of the robotic arm according to the actual operating conditions, and selected angle, load and temperature as the key features, and extracted the stress data under different operating conditions using finite element simulation to form a multidimensional data set. Subsequently, the interference of non-relevant factors was reduced by cleaning the data and filtering the features with Topsis-based entropy weighting method. Finally, a real-time predictive agent model for robotic arm stress was established by the random forest machine learning model. The experimental results show that the R-squared value of the model reaches [X], which exhibits excellent prediction accuracy and stability, and is able to realize the real-time monitoring of the working status of the robotic arm, providing a scientific basis for the safe operation of the robotic arm. This study not only provides a new idea and solution for the design and application of future robotic arms, but also makes an important contribution to improving the safety and intelligence of robotic arms.