Machine learning model outperforms physics-based simulation for hydraulic clutch control

This paper introduces a data-driven method for modeling hydraulic clutch control pressure, addressing nonlinear behaviors like hysteresis and latch transitions. By incorporating current derivative information and using a nonlinear Support Vector Classifier (SVC) to separate operating regimes, the researchers developed a Gaussian Process regression model. This machine-learning approach demonstrated superior accuracy in reproducing measured pressure responses and hysteresis compared to a physics-based Amesim model, suggesting its utility in hardware development and controller calibration. AI

IMPACT This research demonstrates the potential of machine learning models to complement traditional physics-based simulations in complex engineering systems, potentially speeding up development and calibration processes.

RANK_REASON Academic paper detailing a new modeling approach. [lever_c_demoted from research: ic=1 ai=0.7]

Read on arXiv cs.LG →

• hydraulic clutch control
• Simcenter Amesim

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →