Sequence Time Domain Reflectometry (STDR) for Fault Detection on 3-Phase Power Transmission Lines

Final Year Project — Designed by Isaac-Great for a client of Deethunder Nexus

---

Abstract

This project presents the design and simulation of a Sequence Time Domain Reflectometry (STDR) system for detecting and locating faults on 3-phase power transmission lines. STDR injects orthogonal pseudo-noise (PN) sequences simultaneously on all three phases and cross-correlates the reflected signals to determine the fault location.

The system is modelled in MATLAB/Simulink with a custom 3-phase channel model that supports selective fault injection (Line-to-Ground, Short Circuit, Open Circuit) at arbitrary distances. A Python-based post-processing pipeline generates performance tables and publication-quality plots for the results and discussion chapter.

---

System Architecture

The diagram below shows the high-level system overview:

The Simulink implementation of the system block diagram:

---

Project Structure

📦 STDR-3Phase-Fault-Detection
 ├── 📄 STDR_3Phase_Selective.slx       # Main Simulink model
 ├── 📄 STDR.m                          # Batch simulation: Noise × FaultType × Distance sweep
 ├── 📄 selective_stdr.m                # Multi-phase selectivity sweep
 ├── 📄 generatePN3phase.m              # Helper: generates orthogonal 3-phase PN sequences
 ├── 🖼️  System Model.png               # Simulink block diagram screenshot 
 ├── 🖼️  block_diagram.png              # System architecture diagram
 ├── 📄 .gitignore
 └── 📁 analysis/
      ├── 📄 stdr_analysis.py           # Python post-processing & plotting
      ├── 📄 README.md                  # Python pipeline documentation

How It Works

1. PN Sequence Generation (generatePN3phase.m)

Three orthogonal maximal-length PN sequences (length 1023, chip rate 500 kHz) are generated using different generator polynomials:

• Phase A: x¹⁰ + x³ + 1
• Phase B: x¹⁰ + x⁵ + 1
• Phase C: x¹⁰ + x⁷ + 1

Each sequence is upsampled by a factor of 20 and injected simultaneously onto the 3-phase line.

2. Fault Injection & Channel Model (STDR_3Phase_Selective.slx)

The Simulink model contains a Channel_3Phase subsystem that:

• Delays the transmitted signal by Ndelay samples (determined by fault distance and propagation speed v = 2×10⁸ m/s)
• Scales the reflected signal by a reflection coefficient Γ depending on fault type:

Fault Type	Γ
Line-to-Ground (LG)	−0.6
Short Circuit (SC)	−1.0
Open Circuit (OC)	+1.0

• Adds AWGN noise at configurable power levels

3. Correlation & Distance Estimation

Each phase correlates its received signal with its own transmitted PN sequence. The lag at the peak of the cross-correlation gives τ, and the fault distance is recovered as:

d = (τ × v) / 2

4. Selectivity Test (selective_stdr.m)

Only one phase is faulted at a time while the other two remain healthy. The three detectors are observed to confirm that only the faulted phase detector registers a hit — validating cross-phase selectivity.

---

Running the Simulation

Prerequisites

• MATLAB R2021b or later (R2024b recommended)
• Simulink
• Communications Toolbox (required by generatePN3phase.m)

Batch Noise & Distance Sweep

% Open MATLAB, navigate to the project folder, then run:
STDR

Runs 60 simulation scenarios (3 noise levels × 4 fault types × 5 distances) and saves:

• STDR_Noise_Sweep_Results.csv
• STDR_Error_AllPhases.png
• STDR_Linearity_AllPhases.png

Selectivity Sweep

selective_stdr

Runs 27 scenarios (3 faulted phases × 3 fault types × 3 distances) and saves:

• STDR_Full_Selectivity_Study.csv
• STDR_CrossPhase_Selectivity.png

---

Python Post-Processing

After running the MATLAB simulation, use the Python analysis pipeline to generate thesis-quality tables and figures. See analysis/README.md for full instructions.

cd analysis
pip install pandas numpy matplotlib seaborn
python stdr_analysis.py

---

Key Results

Condition	Outcome
Ideal (no noise)	✅ 100% accuracy across all fault types and distances (300 m – 2700 m)
Low noise (σ² = 10⁻⁶)	✅ Negligible accuracy degradation
High noise (σ² = 10⁻⁴)	⚠️ Increased error, especially at longer distances
Cross-phase selectivity	✅ Each detector responds exclusively to faults on its own phase

---

Fault Types Studied

Code	Fault Type	Description
LG	Line-to-Ground	Single phase shorted to ground
SC	Short Circuit	Phase-to-phase short
OC	Open Circuit	Broken/disconnected conductor
LL	Line-to-Line	Phase-to-phase fault (noise sweep only)

---

Simulation Parameters

Parameter	Value
PN Sequence Length	1023 chips
Chip Rate	500 kHz
Samples per Chip	20
Sampling Frequency	10 MHz
Signal Propagation Speed	2 × 10⁸ m/s
Fault Distances Tested	300, 750, 1500, 2250, 2700 m
Noise Levels Tested	0 (ideal), 10⁻⁶, 10⁻⁴

---

Project Credits

This project was done by Deethunder Nexus ventures.

• Website: www.deethundernexus.org
• Email: info@deethundernexus.org

---

License

Proprietary / All Rights Reserved

This project was developed as an academic final-year submission. All intellectual property rights are strictly reserved by the respective institution and the student developer.

Copyright & Usage Restrictions

• No Reproduction or Legal Use: This codebase, models, and associated documentation may not be copied, reproduced, distributed, or repurposed for any commercial, non-commercial, or legal use.
• Client Confidentiality: Due to strict client confidentiality privileges, any request for reproduction, reference, or access to proprietary assets must be formally directed to the organization.

For permissions and inquiries, please contact Deethunder Nexus ventures prior to any intended usage.