RCA Analysis of Multi-Source Faults in Autonomous Driving

This strategy is proposed to address the safety issues caused by inaccurate fault localization and unclear boundary delineation of fault hazard due to the coupling and correlation of multi-source heterogeneous faults in autonomous driving. Construct a multi-source fault entity model using a knowledge graph, locally interpret the characteristics of each fault attribute, and integrate asynchronous message queues and linked data object encoding to handle nested relationships between fault entities; Further utilize the Analytic Hierarchy Process to establish a fault structure evaluation model and evaluation criteria, relying on the relationship chains of various entities in the graph, to quantitatively calculate the correlation weights between faults using the judgment matrix and super matrix. The testing verified the various fault relationship chains and ultimately explained the coupling correlation of each entity, and finally decoupled the key causal factors of multi-source faults, providing a fundamental basis for root tracing of multi-source faults.