Functional Safety Design for Autonomous Vehicle Brake System With Machine Learning Technique

Brake by wire is emerging as a superior alternative to traditional mechanical and hydraulic systems due to its faster response and higher accuracy. The increasing integration of electronic components in modern vehicles necessitates robust fail-operational designs to ensure functional safety. This article presents a novel approach to enhance the reliability and safety of brake by wire systems thorough system analysis, development, and testing. We conduct a system safety analysis to identify potential failure modes inherent in brake by wire systems. For brake pressure estimation and sensor diagnosis, we leverage the sequential data processing capabilities of long short-term memory (LSTM) networks. This data-driven method predicts brake pressure in real-time with high accuracy under various operational conditions. To address pressure sensor failures, we implement a fail-operational control algorithm that adjusts brake pressure based on LSTM predictions. The system is designed with a strong focus on meeting stringent vehicle functional safety requirements. Finally, hardware-in-the-loop tests are conducted to verify safety requirements and the fail-operational control strategy. Fault injection test results demonstrate that the proposed method significantly improves brake system safety, particularly in the event of key brake pressure sensor failures.