Fault Localization in Automotive Power Nets for Utilization in Energy Management Systems Used for Autonomous Driving Based on Graph Theory

The rapid development and continuous improvement of autonomous driving functions poses a number of new challenges to conventional automotive power nets. In addition to the expected increase in consumer power, it must be ensured that energy and power are available in all operating states possible and also in the event of a fault in individual components. This is enabled by novel power net topologies such as ring structured power nets. This has resulted in a constant increase in the complexity of on-board networks, which raises new challenges for modern energy management systems. The application of novel technologies such as semiconductor fuses and the associated high number of degrees of freedom introduce additional challenges. For example, in the event of a fault such as a line short circuit, semiconductor fuses can be opened specifically according to the specifications of the energy management system. This requires the fault to be localized precisely so that suitable countermeasures can be taken. This paper therefore presents a method for successful fault localization using the example of the short-circuit fault by means of graph theory and various simulations. Since fault localization is always accompanied by appropriate fault handling, the link to a Multi-Domain Management is presented, which initiates further steps and adjusts the operational strategy if necessary.