StressNet

Overview

StressNet was built at Larsen & Toubro during my Intelligent Systems internship. It's an end-to-end IoT and machine-learning system for monitoring and predicting structural stress on tourist bridges.

How it works

1. Sensing: an ESP32 reads an HX711 load cell, a DHT22 for temperature and a BMP180 for pressure.
2. Telemetry: readings go to ThingSpeak for logging.
3. Prediction: a Python service polls ThingSpeak and runs an XGBoost classifier that labels each reading.
4. Alerts: live values and predictions go to a Blynk mobile dashboard, with push and email alerts for warning and critical states.
5. Analysis: Power BI is used to review historical data.

The classifier was trained on a synthetic dataset of 1,000+ records, generated from structural rules because real failure data isn't available.

Stack

ESP32, Arduino, Python, XGBoost, ThingSpeak, Blynk, Power BI