Conference article

Driven by the rapid development towards the concept of more electric aircraft in recent years, the power-by-wire technology, especially electromechanical actuators (EMA), is being progressively adopted with the promise of bringing performance improvements for future aircraft in different areas such as weight, maintainability and fuel consumption. However, for safety-critical applications like primary flight controls, the use of EMA introduces specific issues related to the actuator’s response to failure. To that end, focusing on EMA fault diagnosis, this work proposes the development of a model-based quantitative Fault Detection and Isolation (FDI) methodology based on bond graph models. Firstly, an EMA behavioural model is created in bond graph, considering the actuator’s most common failure modes: stator winding faults, backlash, jamming, and mechanical disconnection. Next, this behavioural model is applied to create a Diagnostic Bond Graph (DBG) model, which is based on the use of virtual residual detectors, or residual sinks, for numerical residuals generation, without the need for deriving symbolic Analytical Redundancy Relations (ARRs). These residuals are then evaluated with experimentally defined thresholds for fault detection, indicating if the system behaviour has departed from the acceptable operating range. Finally, the causal and structural properties of the bond graph are used to derive a Fault Signature Matrix (FSM) through the analysis of the DBG model causal paths. The FSM is employed for fault isolation, reducing the number of fault candidates, in order to isolate the component responsible for the faulty behaviour. Simulation results, using bond graph models implemented in the 20-sim environment, show the successful detection and isolation of all EMA modelled faults using the proposed method.

Bond Graph, electromechanical actuator, fault detection, fault isolation