Digital Design and Optimization of Intelligent Parameter-Driven Lubrication Systems

In response to limited multi-condition simulation accuracy and inefficient manual calibration of gas turbine lubrication systems, this paper proposes a hybrid modeling framework that combines physical mechanism models with data-driven techniques. An object-oriented, modular approach is used to develop a multi-physics bearing chamber model encompassing bearing dynamic lubrication, oil–gas two-phase flow, and graphite-seal heat transfer. By coupling the physical model with an intelligent parameter optimization algorithm, hierarchical dynamic corrections are achieved. Through signal-flow modeling and system-level simulation validation, the framework captures parameter uncertainty propagation from components to the system, overcoming the limitations of single-condition calibration. An AI-driven full load spectrum parameter calibration method refines component performance data under maximum-speed ground conditions and typical operating scenarios, reducing the maximum errors of oil pressure by 5%, oil flow by 3%, and oil temperature by 10%, and lowering the overall system simulation error below 5%. Validation results indicate that this approach effectively addresses the generalization shortcomings of traditional mechanism-based models in complex environments, offering a transferable, efficient solution for digital design in aviation power systems.