Hybrid Friction Modeling In Mechanical Systems
Please login to view abstract download link
Friction models are essential for accurately predicting the interactions between sliding surfaces in mechanical systems. Traditional models, such as the Coulomb friction model, offer simplicity by expressing friction by just one constant coefficient. In contrast, more sophisticated models like the Dahl, LuGre, and Stribeck models account for factors such as surface roughness, material properties, and lubrication. These advanced models provide detailed insights into frictional behavior under varying conditions, yet no single model can fully capture the complexities of real-world interactions. In hydraulic systems, friction plays a crucial role in performance. It affects efficiency, energy use, and wear in components like pistons and valves. Accurate friction modeling is essential for optimizing system design and operation. This paper proposes a hybrid approach that combines low-fidelity and high-fidelity friction models within a data-driven framework. Low-fidelity models, with their broad applicability and low computational cost, are integrated with high-fidelity models that offer precise simulations at a higher computational expense. By incorporating empirical data, this hybrid model dynamically adjusts to enhance accuracy and reliability across different operational scenarios. The proposed method aims to balance simulation accuracy with computational efficiency, providing a robust tool for predicting friction in diverse mechanical applications.
