pith. sign in

arxiv: 2604.07281 · v1 · submitted 2026-04-08 · 📡 eess.SY · cs.SY

Active Propeller Fault Detection and Isolation in Multirotors Via Vibration Model

Alessandro Baldini , Riccardo Felicetti , Alessandro Freddi , Andrea Monteri\`u This is my paper

Pith reviewed 2026-05-10 17:31 UTC · model grok-4.3

classification 📡 eess.SY cs.SY
keywords fault detectionfault isolationmultirotorvibration modelpropeller damageIMUactive methodoctarotor
0
0 comments X

The pith

Active perturbation of control inputs allows isolation of propeller blade faults in multirotors using only IMU vibration data.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a method to detect and isolate damage to propeller blades on multirotor aircraft by actively changing the control inputs in a controlled way. This exploits a model of how blade damage produces specific vibrations that can be picked up by the vehicle's inertial measurement unit. Because multirotors have redundant actuators, passive methods struggle to pinpoint which blade is damaged, so the active strategy deliberately excites the system to make isolation possible. The approach is demonstrated through thousands of simulations on an octarotor, showing that time and frequency features from vibrations can distinguish faults without needing extra sensors.

Core claim

By deliberately perturbing the control inputs and analyzing the resulting vibrations captured by the onboard inertial measurement unit through a model that captures blade damage effects, blade faults can be isolated in multirotor vehicles that have significant input redundancy.

What carries the argument

The vibration model that captures the effects of blade damage on the vehicle's dynamics, combined with an active fault isolation strategy that perturbs control inputs to generate distinguishable vibration signatures.

If this is right

  • The method enables fault isolation in highly redundant systems where passive detection is insufficient.
  • Only the existing inertial measurement unit is required, avoiding the need for additional sensors.
  • Both time-domain and frequency-domain features from vibration data can be used to achieve accurate isolation.
  • Large-scale simulation testing on 9600 cases confirms the approach works across varied conditions for an octarotor.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The technique may allow real-time fault mitigation if integrated with control reconfiguration.
  • Similar vibration-based active methods could apply to detecting imbalances in other rotating machinery.
  • Validation on hardware would require confirming that perturbations do not degrade flight performance.

Load-bearing premise

The vibration model accurately represents the effects of real blade damage, and the deliberate input perturbations can isolate faults without compromising vehicle stability or normal flight performance.

What would settle it

Conducting experiments on a physical octarotor with known blade damage where the method incorrectly identifies or fails to isolate the fault when using only IMU data and input perturbations.

Figures

Figures reproduced from arXiv: 2604.07281 by Alessandro Baldini, Alessandro Freddi, Andrea Monteri\`u, Riccardo Felicetti.

Figure 1
Figure 1. Figure 1: NED reference frames for a standard hexarotor. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Pose of the multirotor. 0 10 20 30 40 50 t [s] 300 400 500 600 [r a d = s] _31 _32 _33 _34 _35 _36 _37 _38 [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Octarotor motors angular speeds. During the active FDI the propellers [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Measured linear body acceleration components. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FD residual and FDI residual. The FD residual triggers the active FDI [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Fourier magnitude of measured acceleration before the fault, after the fault, and during each active isolation stage, over the frequency range used for fault [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Fourier magnitude of measured acceleration before the fault, after the fault, and during each active isolation stage, over the frequency range used for active [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: ROC curves for different damping factors d. Motor 1Motor 2Motor 3Motor 4Motor 5Motor 6Motor 7Motor 8No Fault Predicted Class Motor 1 Motor 2 Motor 3 Motor 4 Motor 5 Motor 6 Motor 7 Motor 8 No Fault True Class 16 16 16 16 16 16 16 16 32 (a) d = 5%. Motor 1Motor 2Motor 3Motor 4Motor 5Motor 6Motor 7Motor 8No Fault Predicted Class Motor 1 Motor 2 Motor 3 Motor 4 Motor 5 Motor 6 Motor 7 Motor 8 No Fault True Cl… view at source ↗
Figure 9
Figure 9. Figure 9: Confusion matrix choosing ρFD = 0.0080. but a significant 25% FPR arises. Practically, choosing a large ρFD = 0.0080 is preferable, thus avoiding false positives and accepting the fact that minor faults can go unnoticed. A 1% damping makes the proposed algorithm less effective, as shown in Fig. 8c. If avoiding false positives is prioritized (ρFD = 0.0080), even some of the largest 20% LOE faults can go unn… view at source ↗
Figure 10
Figure 10. Figure 10: Correct FDI ratio w.r.t. the LOE severity and the damping [PITH_FULL_IMAGE:figures/full_fig_p010_10.png] view at source ↗
read the original abstract

In rotary-wing aircraft, rotating blades are exposed to collisions and subsequent damage. The detection and isolation of blade damage constitute the first step in fault mitigation; however, they are particularly challenging when considerable input redundancy is available, as in the case of multirotors. In this article, we propose an active model-based approach that deliberately perturbs the control inputs to isolate blade faults in multirotor vehicles. By exploiting a model that captures the vibrations caused by blade damage, the isolation method relies solely on vibration data from the onboard inertial measurement unit. The strategy is tested in simulation using an octarotor platform, and both time-domain and frequency-domain features are analyzed. Several accuracy-related metrics of the technique are evaluated on a set of 9600 simulations and compared with the most relevant variables.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper proposes an active model-based fault detection and isolation (FDI) strategy for propeller blade damage in multirotor vehicles. It deliberately perturbs control inputs to excite vibrations captured by a damage-specific model and performs isolation using only onboard IMU vibration measurements. The method is evaluated exclusively in simulation on an octarotor platform across 9600 cases, with analysis of time- and frequency-domain features and associated accuracy metrics.

Significance. If the underlying vibration model accurately represents physical blade damage effects, the approach could enable reliable FDI in highly redundant multirotor systems without requiring additional sensors beyond the standard IMU. The scale of the simulation campaign (9600 cases) allows statistical assessment of performance across operating conditions, which strengthens the empirical component if the model fidelity holds.

major comments (3)
  1. [Simulation Results] Simulation Results section: All 9600 evaluations and accuracy metrics rely on the proposed vibration model without any reported comparison to experimental data from physically damaged blades or hardware tests. This leaves the central claim of reliable isolation dependent on an unverified assumption that the simulated effects (mass imbalance, lift loss, etc.) match real aeroelastic and sensor behavior.
  2. [Vibration Model] Vibration Model section: The manuscript provides insufficient detail on the derivation of the vibration model or its validation against real damage mechanisms. Without this, it is impossible to assess whether omitted effects (e.g., blade flexibility, turbulence) would degrade the reported time- and frequency-domain isolation performance.
  3. [Perturbation Strategy] Perturbation Strategy subsection: No analysis or results are presented demonstrating that the deliberate control-input perturbations preserve closed-loop stability and do not compromise normal flight performance, which is a load-bearing assumption for the active FDI approach to be practical.
minor comments (2)
  1. [Abstract] Abstract: The phrase 'compared with the most relevant variables' is vague; the manuscript should explicitly state which variables are used for comparison and the nature of the comparison.
  2. Notation throughout: Several symbols for vibration features and perturbation signals are introduced without a consolidated table or clear cross-references, reducing readability.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback. We address each major comment below, indicating where revisions will be made to the manuscript.

read point-by-point responses

  1. Referee: [Simulation Results] Simulation Results section: All 9600 evaluations and accuracy metrics rely on the proposed vibration model without any reported comparison to experimental data from physically damaged blades or hardware tests. This leaves the central claim of reliable isolation dependent on an unverified assumption that the simulated effects (mass imbalance, lift loss, etc.) match real aeroelastic and sensor behavior.

    Authors: We acknowledge that the study is simulation-based and does not include hardware experiments with physically damaged blades. The 9600 cases enable statistical evaluation under repeatable conditions using a physics-derived model of mass imbalance and lift loss. In revision, we will add an explicit discussion of model assumptions, limitations, and the need for future experimental validation to support the claims. revision: partial

  2. Referee: [Vibration Model] Vibration Model section: The manuscript provides insufficient detail on the derivation of the vibration model or its validation against real damage mechanisms. Without this, it is impossible to assess whether omitted effects (e.g., blade flexibility, turbulence) would degrade the reported time- and frequency-domain isolation performance.

    Authors: We will expand the Vibration Model section with a step-by-step derivation of the vibration equations, including how damage parameters enter the model. We will also add discussion of modeling assumptions and the potential influence of omitted effects such as blade flexibility and turbulence on the reported isolation metrics. revision: yes

  3. Referee: [Perturbation Strategy] Perturbation Strategy subsection: No analysis or results are presented demonstrating that the deliberate control-input perturbations preserve closed-loop stability and do not compromise normal flight performance, which is a load-bearing assumption for the active FDI approach to be practical.

    Authors: We agree that stability and performance impact must be demonstrated. The perturbations are designed to be small and brief. In the revised manuscript, we will include simulation results quantifying closed-loop stability margins and the effect of the perturbations on flight performance metrics such as attitude and position tracking errors. revision: yes

Circularity Check

0 steps flagged

No circularity detected; derivation is self-contained

full rationale

The paper presents a model-based active fault isolation method that perturbs control inputs and extracts time- and frequency-domain features from IMU vibration signals using a vibration model of blade damage effects. All 9600 evaluations are performed in simulation on an octarotor; the approach is described as exploiting physical vibration phenomena rather than any fitted parameters, self-referential definitions, or load-bearing self-citations. No step in the abstract or described chain reduces a prediction or isolation result to its own inputs by construction, so the derivation remains independent of the target claims.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract provides insufficient detail to identify specific free parameters, axioms, or invented entities; the vibration model is referenced but not decomposed.

pith-pipeline@v0.9.0 · 5439 in / 1111 out tokens · 122482 ms · 2026-05-10T17:31:15.945879+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

7 extracted references · 7 canonical work pages

  1. [1]

    Enhancing imu accuracy in mravs: A theoretical and experimental approach to vibration damping, in: 2025 International Conference on Unmanned Aircraft Systems (ICUAS), IEEE. pp. 847–853. Baldini, A., D’Alleva, L., Felicetti, R., Ferracuti, F., Freddi, A., Monteriù, A., 2023a. Uav-fd: a dataset for actuator fault detection in multirotor drones, in: 2023 Int...

    work page 2025

  2. [2]

    Disturbance observer based fault tolerant control for tilted hexarotors, in: 2021 International Conference on Un- manned Aircraft Systems (ICUAS), IEEE. pp. 20–27. Bangura, M., Melega, M., Naldi, R., Mahony, R.,

    work page 2021

  3. [3]

    Aerodynamics of Rotor Blades for Quadrotors

    Aero- dynamics of rotor blades for quadrotors. arXiv preprint arXiv:1601.00733 . Bondyra, A., Gasior, P., Gardecki, S., Kasi´nski, A.,

    work page Pith review arXiv

  4. [4]

    Fault diagnosis and condition monitoring of uav rotor using signal processing, in: 2017 Signal Processing: Algorithms, Archi- tectures, Arrangements, and Applications (SPA), pp. 233–

    work page 2017

  5. [5]

    Vibration-based propeller fault diagnosis for multicopters, in: 2018 International Con- ference on Unmanned Aircraft Systems (ICUAS), IEEE. pp. 1041–1047. Heirung, T.A.N., Mesbah, A.,

    work page 2018

  6. [6]

    Neural network-based propeller damage detection for multirotors, in: 2023 International Conference on Unmanned Aircraft Systems (ICUAS), pp. 17–23. 11 Puchalski, R., Bondyra, A., Giernacki, W., Zhang, Y .,

    work page 2023

  7. [7]

    Actuator fault detection and isolation system for multirotor unmanned aerial vehicles, in: 2022 26th International Con- ference on Methods and Models in Automation and Robotics (MMAR), pp. 364–369. Puchalski, R., Ha, Q., Giernacki, W., Nguyen, H.A.D., Nguyen, L.V .,

    work page 2022