Return
Understanding thermodynamic mechanisms of 6-DOF thermal errors in complex manufacturing feed systems via a novel interpretable network with learned activation functions
J
L
X
T
N
DOI:10.1016/j.ijmachtools.2026.104390.png)
Abstract
En 中文
• Complex spatial–temporal error variations in feed systems are systematically modeled. • Interpretable neural networks decouple thermal drift and dynamic hysteresis. • The model identifies physically meaningful functional relationships from data. • Thermodynamic constraints ensure physically plausible predictive behaviors. • High accuracy is achieved without relying on opaque data-driven correlations.
Keywords:
thermal errors
feed systems
interpretable neural networks
thermodynamic constraints
spatial–temporal modeling
Journal
IF:
18.8
Papers:
3.6K
Citations:
1.8W
