Integrating Optimization and Model Fidelity Selection Methods for Real-Time Driver-in-the-Loop Vehicle Simulators

Automotive OEMs perform extensive prototype testing to configure vehicles for objective criteria (performance), and subjective criteria (handling and comfort). To reduce testing time and costs, OEMs rely on real-time Driver-In-the-Loop Simulators (DIL) running complex Multi-Body Dynamics (MBD) models. Recent advances in simulation technology have increased model accuracy but also operating costs, possibly limiting the viability of real-time DIL applications. Running high fidelity MBD models in real-time is computationally intensive and often requires re-configuration, CAE model de-contenting, solver setting optimization, which can introduce significant analysis errors. This presents a core challenge: selecting model fidelity levels that result in computationally efficient simulations, while maintaining sufficient predictive accuracy. This study introduces a methodology that integrates optimization algorithms with decision-making techniques to select the right fidelity within a combinatorial space of MBD model configurations. A single-objective genetic optimization algorithm identifies compliance settings that reduce Real-Time Factor (RTF). Suitable configurations are evaluated in fully executed dynamic simulations, with accuracy quantified against a baseline model using Root Mean Squared Error (RMSE) and Dynamic Time Warping (DTW). Sensitivity analysis isolates the impact of each compliance setting on RTF and accuracy. Finally, an interactive parametric Multi-Criteria Decision-Making (MCDM) dashboard allows for vehicle development stakeholders to obtain ranked model fidelity alternatives by assigning importance weights for each metric. By integrating decision-making techniques with optimization algorithms, reductions in RTF of over 10% were observed with minimal accuracy drops. The method is demonstrated by using a commercial MBD library, for vehicle models performing a J-Turn and a single-bump maneuver. Fidelity alternatives were generated by varying compliance settings, adjusting the components between rigid and flexible connections. This work delivers a transparent and repeatable decision-making process for balancing accuracy and cost through combining fidelity modulation, optimization, and sensitivity analysis within an interactive platform that incorporates vehicle development stakeholder input.