learning_research_materials.md

Learning and Research Materials

Use this roadmap to understand the project deeply enough to explain it in interviews, reports, and research discussions.

For handwritten revision notes, use:

• docs/handwritten_notes_guide.md

Learning Path

1. ESP32 and Embedded Systems

Learn:

• ESP32 GPIO basics
• I2C communication
• WiFi connection flow
• Arduino vs ESP-IDF
• Serial debugging
• Timing with millis()

Practice:

• Blink LED.
• Read DHT22 only.
• Read INA219 only.
• Print voltage/current/power to serial monitor.
• Then combine both sensors.

Key reference:

• Espressif ESP32-DevKitC documentation: https://docs.espressif.com/projects/esp-dev-kits/en/latest/esp32/esp32-devkitc/user_guide.html

2. Sensor Measurement

Learn:

• Difference between voltage, current, and power.
• Why current sensors must be placed in series.
• High-side current sensing.
• Sensor noise and calibration.
• Why bad grounding causes unstable readings.

Key references:

• Adafruit INA219 guide: https://learn.adafruit.com/adafruit-ina219-current-sensor-breakout/overview
• Adafruit DHT temperature/humidity guide: https://learn.adafruit.com/adafruit-io-basics-temperature-and-humidity/arduino-wiring

3. Renewable-Energy Monitoring

Learn:

• Solar panel I-V behavior.
• Open-circuit voltage.
• Short-circuit current.
• Effect of shading on current and power.
• Effect of temperature on PV performance.
• Why delayed fault detection reduces energy yield.

Practice:

• Plot voltage, current, and power under different loads.
• Compare normal operation against open-circuit behavior.
• Write notes for each fault scenario.

4. IoT and MQTT

Learn:

• MQTT publisher, broker, subscriber.
• Topic structure.
• JSON telemetry payloads.
• QoS levels.
• Local vs cloud data flow.

Practice:

• Publish test MQTT messages from ESP32.
• Subscribe from terminal or Python.
• Log MQTT payloads into SQLite.

Key reference:

• MQTT official site: https://mqtt.org/

5. Data Logging and Analytics

Learn:

• SQLite schema design.
• CSV export.
• Dataset labels.
• Train/test split.
• Confusion matrix.
• False positives and false negatives.

Practice:

• Collect 500 records.
• Export CSV.
• Plot power trends.
• Manually inspect fault labels.

6. Machine Learning and TinyML

Learn:

• Rule-based detection.
• Statistical thresholds.
• Decision Trees and Random Forests.
• Isolation Forest for anomaly detection.
• Model size, RAM use, flash use, inference latency.
• Why edge inference matters when cloud access is unavailable.

Practice:

• Train on synthetic data first.
• Replace synthetic data with real ESP32 data.
• Compare rule-based and ML metrics.
• Export a simple edge model to C.

Key reference:

• TensorFlow Lite Micro paper: https://arxiv.org/abs/2010.08678

7. MATLAB / Simulation Validation

Learn:

• Simple PV equivalent behavior.
• Load changes.
• Voltage drop.
• Fault-condition simulation.
• Difference between ideal simulation and noisy hardware.

Practice:

• Simulate voltage and current trends.
• Compare with hardware plots.
• Document why they differ.

Research Papers and Technical Reading

Start with review papers first. They will give you vocabulary and help position your project.

1. Photovoltaic system fault detection techniques: a review

   • Open-access review of PV fault detection methods.
   • Link: https://link.springer.com/article/10.1007/s00521-023-09041-7

2. Detection, classification, and localization of faults and failures in photovoltaic arrays

   • Recent open-access review focused on fault detection, classification, and localization.
   • Link: https://link.springer.com/article/10.1007/s43937-026-00136-5

3. Methods of photovoltaic fault detection and classification: A review

   • Useful for comparing electrical, thermal, visual, and data-driven methods.
   • Link: https://www.sciencedirect.com/science/article/pii/S2352484722008022

4. Machine learning in photovoltaic systems: A review

   • Good for understanding where ML fits in PV forecasting, control, and fault diagnosis.
   • Link: https://www.sciencedirect.com/science/article/pii/S0960148122009454

5. NREL PV reliability and performance research

   • Good industry/research context for why reliability and long-term monitoring matter.
   • Link: https://www.nrel.gov/pv/reliability-engineering

6. IEA PVPS Task 13 report on advanced algorithms in PV failure monitoring

   • Useful for research positioning and real-world monitoring language.
   • Link: https://www.iea-pvps.org/wp-content/uploads/2021/10/Final-Report-IEA-PVPS-T13-19_2021_PV-Failure-Monitoring.pdf

What You Should Be Able To Explain

After studying and building the hardware, you should be able to answer:

• Why did you choose ESP32?
• Why INA219 instead of only voltage measurement?
• Why use MQTT?
• What does open circuit look like electrically?
• Why does partial shading mainly reduce current?
• What is the difference between rule-based detection and ML detection?
• What is a false positive?
• What is a false negative?
• Why is TinyML useful here?
• What are the limits of your prototype?

Suggested 14-Day Study Plan

Day	Focus	Output
1	ESP32 basics	LED blink and serial monitor
2	DHT22	Temperature readings
3	INA219	Voltage/current/power readings
4	Combined sensors	Serial telemetry
5	MQTT	ESP32 publishes JSON
6	Python logger	SQLite records created
7	Dashboard	Live values visible
8	Normal dataset	1-hour stable run
9	Open circuit and shading	Labeled fault records
10	Temperature/current faults	More labeled records
11	Rule analytics	Metrics and confusion matrix
12	ML model	Model trained on real data
13	TinyML export	ESP32-ready model header
14	Research write-up	Results and discussion updated