Thermal Fault Detector

A signal analysis tool that detects and visualizes electrical faults in power system recordings using thermal color heatmaps. It processes time-series current (I1) and voltage (V1) data, classifies each timestep into a 6-level thermal severity scale, and produces a NVA (Normalized Vibrational Activity) score and a 4-bit binary fault summary.

---

Features

• Thermal color heatmaps for current, voltage, and a fused channel
• 6-level severity scale: Violet → Blue → Green → Yellow → Orange → Red
• Fault-gradient coloring for current: smooth Yellow→Orange→Red within the fault zone
• ROC-only voltage coloring: marks transition instants only — stable periods (whether at nominal voltage or 0 V) show Violet/Blue
• State machine fault tracking with configurable cooldown to prevent premature fault-exit
• NVA score (0.0 = perfect, 1.0 = total fault) with five named health states
• Binary output [COLD | NORMAL | WARNING | DANGER] with human-readable interpretation for all 16 combinations
• Auto-detect file loader: handles .xlsx, .xls, CSV (comma/semicolon), and TSV

---

Input Format

The tool expects a spreadsheet or CSV file with at minimum these three columns:

Column	Description
simOut	Time axis (seconds)
V1	Voltage signal (Volts)
I1	Current signal (Amperes)

Extra columns are ignored. Set FILE_PATH in Cell 2 to point to your data file.

---

Outputs

Output	Description
Current heatmap	Fault-gradient coloring with ROC spikes
Voltage heatmap	ROC-only coloring, marks onset/offset transitions
Fused heatmap	max(current_color, voltage_color) per timestep
Color % breakdown	Time spent at each severity level
NVA score	Weighted scalar health index
Binary string	4-bit dominant-state summary with interpretation

---

Color Scale

Index	Color	Meaning
0	Violet	Cold / at rest
1	Blue	Very low activity
2	Green	Moderate activity
3	Yellow	Elevated / fault entry zone
4	Orange	Active fault — lower severity
5	Red	Active fault — peak severity

---

NVA Score Scale

0.00 – 0.20  →  GOOD
0.20 – 0.40  →  OK
0.40 – 0.60  →  AVERAGE
0.60 – 0.80  →  FAULTY
0.80 – 1.00  →  SEVERE FAULT

---

Key Parameters

Tunable constants defined at the top of Cell 1:

Parameter	Default	Description
REST_VALUE	120.0	Idle signal level for both I1 and V1
REST_TOL	0.5	Tolerance for rest detection (±V/A)
FAULT_COOLDOWN	50	Steps below 2× normal required to exit fault state
HEIGHT	30	Pixel height of heatmap bar

Thresholds computed automatically from the data:

Threshold	Description
normal_I	Median of the lower 50th percentile of active (non-rest) current samples
thr_5x	Fault entry threshold — normal_I × 5
thr_2x	Fault exit candidate — normal_I × 2
fault_ceiling	99.9th percentile of absolute current (prevents single-spike distortion)

---

How It Works

Current Coloring

Two zones are handled separately:

• Normal zone: color is the higher of ROC-based color and level-based color (Blue/Green/Yellow depending on how many multiples of normal_I the current is)
• Fault zone (current > thr_5x): color is the higher of ROC spike color and a fault-gradient color. The fault-gradient maps the current's position between thr_5x and fault_ceiling linearly to Yellow→Orange→Red, producing a natural severity gradient instead of locking the entire fault region to solid Red

A state machine governs fault entry and exit. Once entered, the fault state persists until current stays below thr_2x for FAULT_COOLDOWN consecutive steps.

Voltage Coloring

Voltage uses ROC-only coloring. Rapid transitions (fault onset, recovery) produce Orange/Red spikes; stable periods at any voltage level show Violet/Blue. This prevents a sustained voltage drop from flooding the heatmap with red.

Fused Heatmap

At each timestep, the fused color is max(current_color, voltage_color). Current drives the sustained fault color; voltage contributes only at transition moments.

---

Usage

1. Open the notebook in Jupyter or Google Colab.
2. Set FILE_PATH in Cell 2 to your data file path.
3. Run all cells in order (Cells 1 → 11).

jupyter notebook ThermalFaultDetector.ipynb

---

Requirements

See requirements.txt. Install with:

pip install -r requirements.txt

---

License

Copyright (c) 2026 VEDICWARRIOR. All Rights Reserved. This project is not open source. No permission is granted to use, copy, modify, or distribute this software without explicit written consent from the author. See LICENSE for full terms.