Decentralized Most Permissive Observer Architecture for Brake Leakage Diagnosis in Automotive Systems

Brake systems represent one of the most critical components contributing to the safety of modern automotive systems. They are commonly referenced as one of the most vehicle-related factors in fatal crashes. Usually, the faults in this type of component are provoked by brake fluid leaks affecting the performance and response of the entire system, which require manual maintenance procedures to locate the source of the problem. This study presents a novel Discrete Event System (DES) model capable of detecting brake fluid leakage faults using a decentralized Most Permissive Observer (MPO) architecture for fault diagnosis. The developments were verified via an enhanced virtual environment employing a simulation model of the braking system using Simscape and assessing the event-driven fault diagnosis scheme using Stateflow. A ‘Design of Experiment’ method is performed using a factorial design of 3^k, where k is the number of levels of a factor in the factorial experiment (in this paper k is 4), to conduct the assessment of test cases that combined edge values. The efficacy of the proposed DES model for fault diagnosis is demonstrated across a wide range of simulated scenarios, including fault locations, pressure drops, and single and multiple fault combinations. The proposed method consistently diagnoses the injected fault conditions, showcasing its efficiency under diverse operational conditions of the braking system.