Barbaro, Mario (2025) Development of real-time capable approaches for vehicle simulation and virtual sensing. [Tesi di dottorato]

	Documento PDF Mario_Barbaro_38_COMPLETA.pdf Visibile a [TBR] Amministratori dell'archivio Download (26MB) | Richiedi una copia
	Documento PDF Mario_Barbaro_38_PARZIALE.pdf Visibile a [TBR] Amministratori dell'archivio Download (9MB) | Richiedi una copia

Abstract

This thesis presents the development of high-efficiency model-based methodologies for vehicle dynamics simulation and state estimation, aimed at designing physically consistent and computationally efficient algorithms for both driving simulations and on-board applications. The research was structured along two complementary directions: the development of a comprehensive simulation framework and the design of vehicle state estimators, both covering handling and ride scenarios. In the first part, the multiphysical Magic Formula tire model, originally conceived for handling applications, was extended to include tire enveloping behavior and rigid-ring dynamics, enabling accurate reproduction of small-wavelength road excitations and extending its applicability to tire ride intermediate-frequency domain, while maintaining low computational cost and limited parameterization requirements. The formulation was validated through indoor testing and subsequently integrated into a full-vehicle simulation framework, operating in co-simulation with thermal and wear tire models and a multibody vehicle model. An innovative outdoor testing campaign was then designed to perform both handling and ride-oriented tire parameterization using a single vehicle sensor setup. The same dataset was also employed to validate the complete full-vehicle assembly and to analyze accuracy across all the explored operating conditions. The second part focused on model-based state estimation. For handling applications, several Kalman Filter–based sideslip angle estimators were benchmarked to determine the optimal trade-off between accuracy and computational efficiency across different driving conditions. Furthermore, the integration of thermal effects into the estimation algorithm was investigated, demonstrating improved accuracy over a wide thermal range and extending the multiphysical modeling approach, already adopted in simulation scenarios, to virtual sensing applications. For ride applications, a reduced-order full-vehicle observer was developed and validated on simulation data, generated through an high-fidelity multibody model. The estimation results shown reliable reconstruction of vertical dynamics and road profile even under noisy sensor configurations. Although still at an early stage, this formulation, designed to be fully observable and to explicitly account for suspension constraints, provides a robust basis for future experimental validation and estimator enhancement to allow application in real-world scenarios. Overall, the developments presented in this thesis represents the foundation for future testing and validation of virtual sensing logics. The proposed simulation framework will provide a controlled and physically consistent environment to validate and further improve the designed state estimators, enabling performance assessment under various maneuvers, vehicle configurations, and boundary conditions.

Downloads

Actions (login required)

Modifica documento