What Is Fault Tolerant Control?

Fault tolerant control is a branch of control engineering concerned with maintaining acceptable closed-loop performance when a component of a controlled system fails. Control systems for aircraft, industrial plants, or autonomous vehicles contain actuators, sensors, and processors whose failure can degrade performance or destabilize the system entirely. Fault tolerant control designs the feedback architecture, the diagnostic logic, and the recovery strategy so that a confirmed component failure triggers an automatic reconfiguration that sustains the essential control objectives.

The field draws from control theory, signal processing, and reliability engineering. Two broad architectures exist: passive fault tolerant control, in which the controller is designed to be inherently robust to a predefined set of failures without any explicit knowledge of when they occur; and active fault tolerant control, in which a fault detection and diagnosis module monitors the system continuously and triggers a controller redesign or parameter update once a failure is confirmed.

Fault Detection and Diagnosis

The first step in active fault tolerant control is correctly identifying that a fault has occurred, which component has failed, and the magnitude of the failure. Fault detection and diagnosis (FDD) methods include model-based approaches, in which a mathematical model of the healthy system generates predicted outputs that are compared against measured outputs, and the residuals are tested statistically to detect and isolate the failure. Data-driven approaches use historical sensor data to train classifiers or anomaly detectors without requiring an explicit system model. Kalman filter variants, including the extended Kalman filter (EKF) and unscented Kalman filter (UKF), are widely used for FDD in nonlinear systems because they produce minimum-variance estimates of the system state and can be augmented with fault parameters to detect their onset simultaneously with state estimation.

Research on fault-tolerant flight control for sensor and actuator failures in ISA Transactions demonstrates how neural network-based FDD can accommodate simultaneous actuator and sensor faults in flight control without requiring a complete system shutdown.

Controller Reconfiguration

Once the FDD module has confirmed a fault and estimated its character, the control law must be updated to work within the reduced capability of the damaged system. Reconfiguration strategies range from gain-scheduled control, in which a library of pre-computed controllers is switched based on fault type, to model predictive control approaches that replan the input trajectory online using an updated system model that reflects the confirmed failure. Redistribution of control effort across redundant actuators is a common form of reconfiguration: if one hydraulic actuator on an aircraft control surface fails, the remaining actuators accept a larger share of the required force. A review of reconfigurable fault-tolerant control methods in the European Journal of Control surveys the stability guarantees available under each reconfiguration architecture and the assumptions each requires.

Passive Fault Tolerant Control

Passive approaches embed robustness to a defined set of faults directly into the nominal controller design, using H-infinity or linear parameter-varying frameworks that guarantee performance bounds across the entire anticipated fault set without needing a real-time FDD module. The trade-off is conservatism: a passive controller must handle all anticipated faults simultaneously, which typically limits its nominal performance. A survey of passive and active methods in fault-tolerant control from Springer Nature provides a unified framework for comparing the two approaches and identifies the system conditions under which each is preferable.

Applications

Fault tolerant control has applications in a wide range of fields, including:

• Aircraft flight control systems, where actuator or sensor failure must not cause loss of control
• Chemical and petrochemical process plants, maintaining safe operation through instrument failures
• Autonomous underwater and aerial vehicles, sustaining navigation after thruster or sensor faults
• Wind turbines, reconfiguring pitch and torque control around blade actuator failures
• Automotive powertrain and active safety systems under sensor degradation