Multi-sensor Fault Detection and Isolation for EMB System Based on Unknown Input Observer

Abstract

Electro-mechanical brake (EMB) systems generate the braking force with electric motors and use several sensors for feedback control. This paper presents a model-based sensor fault diagnostic approach that employs an Unknown Input Observer (UIO) technique to classify sensors as being healthy or unhealthy and even simultaneously occurred faults in two different sensors. The proposed scheme is organized by UIO logic and a decision-making unit. UIO logic estimates sensor values using the inputs and outputs of EMB system and a decision-making unit identifies the faulty sensor based on a deterministic rule and the estimated ones from UIO logic. Unlike other model-based fault detection and isolation schemes, the proposed one needs only one UIO instead of a bank of observers and UIOI is sufficient for detecting and isolating faults simultaneously occurred. Simulation results are presented to verify the effectiveness of this proposed diagnostic approach.