TRACK: Tool-Augmented Reasoning over Time Series for Fault Diagnosis in Power Management System
Organizations: Lorenz Labs, Analog Devices Inc., USA
Abstract
Fault diagnosis in power management systems running multiple loads relies on experts manually inspecting current and voltage waveforms, which fails to scale across deployed units. We present TRACK, which pairs a time series encoder with a Small Language Model to produce structured, physics-grounded diagnostic reports, calling tools for exact, verifiable calculations. It operates on-device, keeping proprietary sensor data local and avoiding expensive API calls. We introduce a rubric-based framework that evaluates these reports by scoring structure, signal fidelity, physics reasoning, diagnosis, and actionability through per-criterion verdicts. On a 15-class multi-load simulated benchmark, TRACK achieves over 95% fault-category accuracy and generalizes to out-of-distribution faults, whereas the best encoder-probing baseline reaches only 58% accuracy.
Figures & tables
| Model | ID | OOD | Combined |
| Trans-CNN + MLP | 45.53 | 26.10 | 40.16 |
| Chronos + MLP | 56.04 | 16.95 | 45.24 |
| Trans-CNN + MLP with Baseline | 56.58 | 28.95 | 48.95 |
| Chronos + MLP with Baseline | 65.71 | 19.29 | 52.88 |
| Trans-CNN + MLP with Baseline & Load | 63.58 | 29.33 | 54.12 |
| Chronos + MLP with Baseline & Load | 72.22 | 20.29 | 57.87 |
| # | Encoder | Calc. Style | St3 a | St4 b | Frozen c | Accuracy | Consistency | Composite |
|---|---|---|---|---|---|---|---|---|
| 1 | Trans-CNN | Direct generative | ✗ | ✓ | ✗ | 92.70 | 95.52 | 77.98 |
| 2 | Trans-CNN | Equation tags | ✗ | ✓ | ✗ | 93.88 | 96.06 | 85.85 |
| 3 | Trans-CNN | Tool call | ✗ | ✓ | ✗ | 94.12 | 97.76 | 94.15 |
| 4 | Chronos | Tool call | ✗ | ✓ | ✓ | 92.70 | 96.82 | 94.09 |
| 5 | Chronos | Tool call | ✗ | ✓ | ✗ | 94.64 | 95.48 | 94.95 |
| 6 | Chronos | Tool call | ✓ | ✓ | ✓ | 95.42 | 97.97 | 95.01 |
Appendix figures & tables8 assets
Supplementary material from the paper’s appendix.
Appendix
| Load Type | Condition | Train | Test | OOD |
| LED strip | Healthy | 5,600 | 700 | 130 |
| Connector degradation | 800 | 100 | 95 | |
| Insulation damage | 800 | 100 | 90 | |
| LED short circuit | 800 | 100 | 185 | |
| Resistive | Baseline (reference) | 2,800 | 350 | – |
| Cap. ESR (minor) | 406 | 51 | 50 |
| OOD Condition | Load |
|---|---|
| Wrong resistor (5 ) | Resistive load |
| Wrong resistor (20 ) | Resistive load |
| Heatsink missing | Thermal pad |
| Wrong pad type | Thermal pad |
| Thermal near-short | Thermal pad |
| Duty cycle stuck high | PWM load |
| Size | Acc. | Consistency | Composite | Latency (s) |
|---|---|---|---|---|
| 0.5B | 84.8 | 59.3 | 83.5 | 42 ∗ |
| 1.5B | 93.3 | 75.8 | 88.8 | 53 ∗ |
| 3B | 94.6 | 97.5 | 95.1 | 65 |
| # | Category | Criterion | Method | Weight | Penalty |
|---|---|---|---|---|---|
| 1 | Structure | All 4 sections present | code | 1.0 | — |
| 2 | Structure | Valid JSON | code | 1.0 | — |
| 3 | Structure | Within word budget ( 300; headlamp 450) | code | 1.0 | — |
| 4 | Signal Profile | No interpretive language in signal section | code | 1.0 | |
| 5 | Signal Profile | value matches ground truth ( 1%) | code | — | — |
| 6 | Signal Profile | , Droop, match ( 1%, per load) | code | — | — |
Fault Diagnosis
Fault diagnosis identifies, localizes, and explains failures in engineered and computational systems using sensor data, logs, models, and domain knowledge.
See all 157 papers