IoT Bridge Vibration Monitor with Cloud Dashboard

Bridges are inspected visually on schedules, but structural damage — a cracked girder, a loosened bearing — changes how a bridge vibrates long before it is visible. Structural health monitoring (SHM) is the engineering answer: measure the vibration signature, and watch for the modal frequencies drifting as stiffness is lost. Full SHM installations cost lakhs, so students study the theory without ever touching it. This project is a working, classroom-scale SHM loop. An accelerometer node on a model bridge girder samples at 100 Hz and streams to the cloud; the pipeline computes RMS acceleration, runs an FFT, picks the modal peaks, and compares each mode against the design estimate. A 30-day drift chart makes the core SHM principle tangible: healthy structure, flat line; stiffness loss, downward drift past the alert band. The dashboard also logs events — heavy-vehicle passes, wind gusts — so the student learns to separate normal excitation from structural change, which is the actual skill in vibration monitoring.

1. The MPU6050 samples acceleration at 100 Hz; the ESP32 batches and streams the data to the cloud over Wi-Fi.
2. The cloud pipeline computes RMS acceleration over rolling 10-minute windows and checks it against the 0.15 g threshold.
3. An FFT runs on each data window; peak detection identifies the modal frequencies (bending and torsional modes).
4. Measured modes are compared against the design estimates from the bridge model's documented calculation, giving per-mode drift percentages.
5. The dashboard renders the live waveform, the FFT spectrum with marked peaks, and the modal comparison table.
6. Daily baseline FFTs are stored; the 30-day drift chart plots 1st-mode drift against the ±2% healthy band.
7. Drift beyond ±5% (sustained) would raise an inspection alert — the threshold logic is documented for the student's own structure.

Tech stack:

• ESP32 + MPU6050 accelerometer
• 100 Hz sampling firmware
• Cloud ingestion (MQTT/HTTP design)
• FFT spectral analysis
• Time-series storage
• Cloud dashboard (HTML/CSS/JS + canvas charts)

• [100 Hz vibration sampling] MPU6050 accelerometer on ESP32, streaming 3-axis acceleration to the cloud in near real time.
• [Live time-domain view] Scrolling acceleration waveform with RMS computed over 10-minute windows against a 0.15 g alert threshold.
• [FFT spectrum] Frequency-domain view with the dominant modal peaks automatically marked (1st/2nd bending, torsion).
• [Modal comparison table] Measured vs design frequencies per mode with percentage drift and status pills.
• [30-day drift chart] Long-term 1st-mode tracking inside a ±2% healthy band — the damage indicator, demonstrated honestly.
• [Event log] Heavy-vehicle passes, wind gusts and baseline uploads timestamped alongside vibration data.
• [Baseline procedure] Buyer-run method to establish the healthy signature of their own model bridge before drift monitoring begins.
• [Cloud pipeline] Documented path from sensor to dashboard: sampling, transmission, FFT, storage, visualisation.

• Accelerometer node prototype (ESP32 + MPU6050 on model bridge mount)
• Sampling and streaming firmware
• Cloud dashboard web app (time domain, FFT, modal table, drift chart)
• Cloud pipeline documentation
• Model bridge design notes with modal estimates
• Baseline calibration procedure
• Project report PDF (background, vibration theory, FFT, SHM principles, methodology)
• PPT presentation for final review
• Viva Q&A preparation document (modal analysis, FFT, aliasing, drift as damage indicator)

• Built for a model bridge; the modal frequencies are estimates for the model, and the report never claims they represent a real bridge.
• 100 Hz sampling resolves modes up to 50 Hz (Nyquist) — adequate for the model, not for full-scale structures.
• Drift-as-damage is demonstrated as a principle; the demo structure is not actually damaged during the project.
• Wi-Fi streaming needs local network coverage at the installation point.
• Temperature effects on the model (stiffness changes with heat) can mimic drift — the report discusses this confounding factor honestly.