SM-Bello/PHI-CTRL-F16-Fault-Recovery-Telemetry

PHI-CTRL F-16 Actuator Fault Recovery Dataset

High-Fidelity JSBSim 6-DOF Telemetry for Physics-Hybrid Self-Healing Flight Control

Official verification artifacts of the PHI-CTRL (Physics-Hybrid Integrity Control) architecture — a digital-twin-driven, self-healing flight control framework that actively compensates actuator degradation in real time.

Author: Mohammed Bello Sani (SM-Bello) Affiliation: Air Force Institute of Technology (AFIT), Kaduna · Penelope Inc. / PHI Lab Project Homepage: https://huggingface.co/SM-Bello Related Dataset: ZeroTwin-UAV Synthetic Physics-Informed Multi-UAV Fault Telemetry

---

1. Why this dataset exists (Project Context Map)

Project Component	Status	Location / Notes
PHI-CTRL Architecture	Defined	Cascaded PID baseline + exact (1/\gamma) physics compensation + MRAC + Luenberger observer + RL residual
Plant Decision	F-16 (mature)	Upstream JSBSim c172p is BETA and cannot produce reliable airborne trim
Classical Baseline Gate	PASS	Hold-trim, no-fault, and 20 % elevator effectiveness loss all pass strict criteria
Adaptive Layers (MRAC / Observer / Residual)	In progress	Ready for integration on the proven F-16 plant
Negative Result (c172p)	Documented	Public warning so researchers avoid the same dead-end
Target Paper	Q1 Journal	Architecture + verification pyramid + physics-informed reconfiguration

This dataset is the public, citable foundation of the successful classical baseline on the F-16. All future full-stack (Cases B–E) runs will be added as new configurations.

---

2. Dataset Summary

High-rate (120 Hz) 6-DOF flight telemetry generated with the mature JSBSim F-16A model under three controlled conditions:

Configuration	Description	Fault	Duration	Pass Status
no-fault	Nominal closed-loop altitude/speed hold	None	60 s	PASS
hold-trim	Open-loop trim hold validation	None	60 s	PASS
20pct-elev	20 % elevator effectiveness loss (γ = 0.80) injected at t ≈ 15 s	Elevator	60 s	PASS

Initial Condition (all runs)

• Altitude: 15 000 ft
• Airspeed target: 400 kts (soft)
• Aircraft: General Dynamics F-16A (JSBSim PRODUCTION model)

Pass Criteria Applied

1. Pre-fault |Δh| < 120 ft
2. Full 60 s duration (no bailout)
3. Minimum airspeed ≥ 55 % of target
4. Pitch within ±40°

---

3. File Structure & Contents

phi-ctrl-f16-fault-recovery/
├── no-fault/
│   ├── baseline_recovery_log.csv      # 7200 rows × 18 channels
│   ├── baseline_recovery.png
│   └── console_log.txt
├── hold-trim/
│   ├── baseline_recovery_log.csv
│   ├── baseline_recovery.png
│   └── console_log.txt
├── 20pct-elev/
│   ├── baseline_recovery_log.csv      # Primary fault case
│   ├── baseline_recovery.png
│   └── console_log.txt
└── README.md                          # This file

Telemetry Channels (CSV columns)

Column	Unit	Description
time_s	s	Simulation time
alt_ft	ft	Geodetic altitude
vc_kts	kts	Calibrated airspeed
theta_deg	deg	Pitch attitude
phi_deg	deg	Bank angle
alpha_deg	deg	Angle of attack
hdot_fps	ft/s	Vertical speed
q_dps	deg/s	Pitch rate
throttle	–	Throttle command [0–1]
elev_cmd	–	Elevator command (pre-plant)
elev_plant	–	Elevator position after fault scaling
pitch_trim	–	Pitch trim command
speedbrake	–	Speedbrake command
pitch_cmd_deg	deg	Inner-loop pitch command
alt_err	ft	Altitude tracking error
vel_err	kts	Airspeed tracking error
fault_active	0/1	Fault injection flag
eff_gamma	–	Actuator effectiveness (1.0 → 0.8)

---

4. How the data was generated

• Simulator: JSBSim v1.3.1 (GitHub build)
• Aircraft model: f16 (PRODUCTION release by Erik Hofman)
• Control law: Classical cascaded altitude → pitch + auto-throttle (PHI-CTRL Layer 1 baseline, V19)
• Fault model: Multiplicative elevator effectiveness loss (\gamma = 0.80) applied after settle
• Settle strategy: Path-first (altitude-rate & pitch), frozen elevator/throttle during settle, hard rate limits
• Logging rate: 120 Hz

Full generation scripts and the complete PHI-CTRL clean workspace are available in the accompanying project release.

---

5. Intended Use Cases

• Benchmarking classical vs. adaptive / fault-tolerant controllers
• Training residual RL policies (PPO / SAC) on realistic actuator degradation
• Validating digital-twin health estimators and (\gamma) identification algorithms
• Studying energy management (TECS) under control-surface loss
• Reproducing the PHI-CTRL baseline gate before adding MRAC / observer / residual layers
• Negative-result research: understanding why the JSBSim c172p model fails airborne trim

---

6. Negative Result – JSBSim c172p (Important Warning)

The upstream JSBSim c172p model is marked BETA. Extensive attempts (grid trim, forced IC, engine spool-up windows, gain retuning, TECS, rate limiting) produced:

• Persistent gear-ground contact on every initial condition
• Native / grid trim returning near-zero throttle or sustained descent
• Unusable airborne equilibrium before any fault injection

Conclusion: Do not use c172p as a research plant for adaptive or fault-tolerant control work that requires a reliable classical baseline. Switch to mature models (F-16, etc.). A dedicated negative-result note will be published alongside this dataset.

---

7. Quickstart

from datasets import load_dataset
import pandas as pd

# Load the 20 % elevator fault case
ds = load_dataset("SM-Bello/PHI-CTRL-F16-Fault-Recovery-Telemetry", data_files="20pct-elev/baseline_recovery_log.csv")
df = pd.DataFrame(ds["train"])

print(df[["time_s", "alt_ft", "vc_kts", "eff_gamma", "fault_active"]].describe())

---

8. Citation

@misc{bello2026phictrl-f16,
  author       = {Mohammed Bello Sani},
  title        = {PHI-CTRL F-16 Actuator Fault Recovery Dataset: High-Fidelity JSBSim 6-DOF Telemetry for Physics-Hybrid Self-Healing Flight Control},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/SM-Bello/PHI-CTRL-F16-Fault-Recovery-Telemetry}},
  note         = {Part of the PHI-CTRL (Physics-Hybrid Integrity Control) research program}
}

---

9. License & Contact

• License: CC-BY-4.0
• Author: Mohammed Bello Sani (@SM-Bello) · smbello.vercel.app
• GitHub: Sm-bello
• Enterprise: penelope-inc.vercel.app

For questions, collaboration, or to contribute additional fault cases (asymmetric, high-AoA, multi-surface), open an issue or contact via Hugging Face.