Frequency-Domain Sensitivity Analysis Method for Mitigation of Automotive Power Supply Disturbances

Automated vehicles need fail-operational power supply systems. Safety-relevant functions must be maintained even if a fault, for example a short circuit, occurs in the system. However, disconnecting such a fault creates transient switching pulses that propagate through the system and may affect further components. These cascading faults must be identified during development and prevented by optimizing the system. For this, efficient sensitivity analysis methods are very useful to reveal the most influential system parameters. As in complex power systems a large number of configurations have to be simulated, time-domain simulations are impractical due to the often very long simulation runtimes. An alternative approach in frequency domain is proposed in this contribution. Transfer-function sensitivities are determined, weighted, and combined to approximately describe the required time-domain sensitivities. The method is demonstrated on a large supply system model and validated using accurate time-domain simulations. It proves to be successful in determining sensitive parameters and mitigating transient pulses. Furthermore, a reduction in computing time of more than 99 % could be achieved compared to time-domain calculations.