Virtual prototyping of vehicular electric steering assistance system using co-simulations

Doktorsavhandling, 2021

Virtual prototyping is a practical necessity in vehicle system development. From desktop simulation to track testing, several simulation approaches, such as co-simulation and hardware-in-loop (HIL) simulation, are used. However, due to interfacing problems, the consistency of testing results may not be ensured. Correspondingly, inherent inaccuracies result from numerical coupling error and non-transparent HIL interface, which involves control tracking error, delay error, and attached hardware and noise effects. This work aims to resolve these problems and provide seamless virtual prototypes for vehicle and electric power-assisted steering (EPAS) system development.
The accuracy and stability of explicit parallel co-simulation and HIL simulation are investigated. The imperfect factors propagate in the simulation tools like perturbations, yield inaccuracy, and even instability according to system dynamics. Hence, reducing perturbations (coupling problem) and improving system robustness (architecture problem) are considered.
In the coupling problem, a delay compensation method relying on adaptive filters is developed for real-time simulation. A novel co-simulation coupling method on H-infinity synthesis is developed to improve accuracy for a wide frequency range and achieve low computational cost. In the architecture problem, a force(torque)-velocity coupling approach is employed. The application of a force (torque) variable to a component with considerable impedance, e.g., the steering rack (EPAS motor), yields a small loop gain as well as robust co-simulation and HIL simulation. On a given EPAS HIL system, an interface algorithm is developed for virtually shifting the impedance, thus enhancing system robustness.
The theoretical findings and formulated methods are tested on generic benchmarks and implemented on a vehicle-EPAS engineering case. In addition to the acceleration of simulation speed, accuracy and robustness are also improved. Consequently, consistent testing results and extended validated ranges of virtual prototypes are obtained.